Copyright Registration Guidance: Works containing material generated by AI

federalregister.gov

411 pointsnagonago402 comments

> If a work's traditional elements of authorship were produced by a machine, the work lacks human authorship and the Office will not register it.

> For example, when an AI technology receives solely a prompt from a human and produces complex written, visual, or musical works in response, the “traditional elements of authorship” are determined and executed by the technology—not the human user. Based on the Office's understanding of the generative AI technologies currently available, users do not exercise ultimate creative control over how such systems interpret prompts and generate material. Instead, these prompts function more like instructions to a commissioned artist—they identify what the prompter wishes to have depicted, but the machine determines how those instructions are implemented in its output.

> For example, if a user instructs a text-generating technology to “write a poem about copyright law in the style of William Shakespeare,” she can expect the system to generate text that is recognizable as a poem, mentions copyright, and resembles Shakespeare's style.

> But the technology will decide the rhyming pattern, the words in each line, and the structure of the text.

> When an AI technology determines the expressive elements of its output, the generated material is not the product of human authorship.

> As a result, that material is not protected by copyright and must be disclaimed in a registration application.

This is the crux of the argument for me.

Based on this reasoning, does that mean that you could not claim copyright over a picture captured by a motion sensor camera.

Basically you instructed it to capture when it determined movement and the camera determined how those instructions are implemented in its output.

Has anyone tried to claim copyright over that before? I wouldn't be surprised if it was rejected.

Yes, definitely. Trail/nature cam photos fit this description.

https://en.m.wikipedia.org/wiki/Monkey_selfie_copyright_disp...

I'd think so, considering how security camera installers never put any effort into capturing footage themselves and are usually the only ones to have the opportunity to do so.

Also look at all the post processing and AI that goes into smartphone cameras right now. The human element doesn't even choose which photo this is done algorithmically and the photo is stitched together in lots of different ways. The human just prompted: 'Now!'

…as long as you don’t count choice of subject matter, shooting angle, composition, crop, or moment to shoot to be creative choices.

We don't need to invoke smartphones here. There's no more or less creative input from a human in traditional photography.

Timing is only one aspect of creativity.

In most settings a video camera’s output is copyrighted because of how you select the location and angel to film. Motion capture has all of those elements, and simply doesn’t capture 99.9% of the possible images. ie: Someone picked a log crossing the river as a place where something interesting will happen and they setup the shot to look interesting.

Which is the general argument why such setups fall under copyright. Though as always there are many edge cases.

There was a case last year where a person sued a content creator on YouTube for distributing parts of their accidental livestream. [1] The argument for fair use was that there was no creative input, so fair use. Though, I really do not know, as I think images and videos can be copyrighted exactly for the reason in your comment.

[1] https://www.youtube.com/watch?v=Uk1Tzqc5vk4

How is setting up a motion camera on a log not similar to setting up a prompt for ChatGPT?

The camera doesn’t decide what to put on the picture?

It does… or at least the human doesn’t.. it’s motion activated

Motion capture rigs are doing capture whatever trees/rocks/buildings etc the person setting it up aimed at. The only thing motion capture does is pick interesting times, but you can write software that does the same thing with a video.

The camera is not making a decision. The human made the decision to set the camera up to take pictures when motion activated.

Yeah, I think in the general case there's going to be a significant amount of selection and arrangement of the works from the AI, though I guess that even then none of the underlying images (or such) of the AI get protection.

I wonder if they know about the "inpainting" technique where the human generates an image, erases parts of it, then adds another image and has the AI fill in the blanks and make a new image based on the resulting image?

I think that'd be harder to disentangle the way they do right now to say these parts are copyrightable because the human did them and those parts are not because the AI did them since you can't really separate the inputs at that point.

I would argue that location and angle to film are analogous to specifying that you want a poem (as opposed to an essay) about copyright law (as opposed to quantum mechanics) in the style of William Shakespeare (as opposed to John Milton).

Basically, those are very broad parameters specified, and the machine is the one that decides to produce the output. You aren't even picking the timing or the subject or likely even the focal length. The machine is doing that based on its algorithms (and with autofocus) even the focal length.

It’s not just location and angle but where to aim the motion capture sensor, what if any bait to use etc. There’s effectively infinitely many possible ways to set of these things.

With chat bot’s every possible response is predetermined when the algorithm is setup. Saying you should get copyright is like saying you should own the results of a search engine response to your query.

There is so many places to inject creativity into the process.

Before the capture you have: Chosing the location, posing the camera, adjusting objects in the background, setting up bate to attract wildlife, choosing optimal camera settings. Even selecting a camera could be considered a creative input.

After the capture, the artist can choose which of the thousands of captures best fits their vision, crop it and applying color correction.

I don't think a single one of those actions can make it a creative work on its own. But when you combine a few of them along with intent, it becomes a creative work.

You can do most of that with prompting and tools like control net in stable diffusion as well. And then take it into photoshop and do changes, feed it back into img2img and inpaint until your hearts content. One can spend multitudes more time than it too to tap the shutter button on an iPhone. In Midjourney of course, you have far less control.

There's a ton of control that prompt crafting alone gives you. There's also the choice of which version of midjourney to use, and various meta options that it gives you.

The copyright office clearly has not the slightest clue about what they're talking about when they claim that the AI is the sole creator here. AI generated content has always been a collaboration with humans, and there's always human creativity involved.

You can own the copyright to the prompt used to generate the output.

But the algorithm isn’t collaborating, every possible outcome is fixed when the algorithm is finalized and users can’t actually change the possibilities. I clearly don’t own the copyright to my Google search results even if my query is quite unique.

The essay that has helped me most to think about this sort of thing has been Brian Eno's Composers as Gardeners.[1] It's about music, but I think it applies equally well to AI-generated art, where humans collaborate by writing the algorithms, choosing the subjects, providing the prompts, and curating the results.

Here's an excerpt:

"...essentially the idea there is that one is making a kind of music in the way that one might make a garden. One is carefully constructing seeds, or finding seeds, carefully planting them and then letting them have their life. And that life isn't necessarily exactly what you'd envisaged for them. It's characteristic of the kind of work that I do that I'm really not aware of how the final result is going to look or sound. So in fact, I'm deliberately constructing systems that will put me in the same position as any other member of the audience. I want to be surprised by it as well. And indeed, I often am.

"What this means, really, is a rethinking of one's own position as a creator. You stop thinking of yourself as me, the controller, you the audience, and you start thinking of all of us as the audience, all of us as people enjoying the garden together. Gardener included.

"We're used to the idea, coming from the industrial and very intelligent post-Enlightenment history that we have, we're used to the idea that the great triumph of humans is their ability to control. And indeed, that must be the case, to some extent.

"What we're not so used to is the idea that another great gift we have is the talent to surrender and to cooperate. Cooperation and surrender are actually parts of the same skill. To be able to surrender is to be able to know when to stop trying to control. And to know when to go with things, to be taken along by them. And that's a skill that we actually have to start relearning. Our hubris about our success in terms of being controllers has made us overlook that side of our abilities. So we're so used to dignifying controllers that we forget to dignify surrenderers...

"...my idea about art as gardening is to sort of revivify that discussion and to say let's accept the role of gardener as being equal in dignity to the role of architect, as in fact, is shown in this lovely pavilion here."

[1] - https://www.edge.org/conversation/brian_eno-composers-as-gar...

Simply planting a tree doesn’t give you copyright of the shape the tree ends up in the way you would on a sculpture.

That’s been the case for a very long time, you need significant control over the specific output because it quantifies. A garden is copyrightable based on the layout of the plants when that involved creativity.

That’s been a legal distinction for a very long time, and this statement is simply consistent with that history.

> But the algorithm isn’t collaborating, every possible outcome is fixed when the algorithm is finalized and users can’t actually change the possibilities.

But that actually reinforces the idea that all of the creative work is in the prompt, everything else is purely mechanical process implementing the command given by the prompt. Arguably, its analogous to saying that a programmer can copyright the prompt but not the resulting image is like saying I can copyright source code, but can have no copyright on the output of the compiler.

Compilers don’t create a new copyright the output is covered as a derivative work.

However, derivative works have clear limitations and the output of a chat program doesn’t qualify any more than you own the copyright of what someone says when you interview them.

Put another way you don’t own the copyright on the specific shape of a tree as a sculpture because you selected its species when you planted it.

> Compilers don’t create a new copyright the output is covered as a derivative work.

A derivative work is a separate work that, considered apart from the one it is derived from, separately has the required creative input to be a copyrightable work, and it does, in fact, have a separate copyright from the original (creating derivative works is an exclusive, but licensable, right of the copyright holder of the original, but the copyright of a derivative is separate.)

I don’t know if that distinction was intended as a limitation, “sound recording” is listed as a derivative work in the statute. It also clarifies that “Copies” are material objects, other than phonorecords

So, mechanical transformation such as rendering a webpage at 150% scale is seemingly a derivative work even if there isn’t any creativity in the process.

You could argue the person who set up the camera chose the angle or frame, just like a normal camera operator.

In the same way that the person using ChatGPT chose the prompt?

No.

How are they different?

I disagree, because instructing the device the circumstances to capture the image, whether it's a timer or motion or some other detector or your direct button press, that is not the only element of photography. They're saying the only creative elements of the work were done by machine.

If you asked me to write a poem about copyright law in the style of Shakespeare, is that something you can copyright, or a work derived from something you can copyright? I don't think so.

It's not that producing a prompt for a writer might not be a somewhat creative act itself, but that's not really recognized by copyright law. Ideas aren't copyright, works are. Colloquially people might agree that a work could be significantly derived from an idea, but I don't think that's how copyright law itself works.

Linux was written from scratch, copying no code from Unix and therefore is not a derived work or infringing on Unix copyright. Unquestionably it faithfully copied many ideas verbatim from Unix, the invention or development or discovery of those ideas were probably the most substantial creative contributions that Ritchie and Thompson had, and the code itself relatively mundane (though expertly written) by comparison. Those ideas/inventions are not classed as copyright works though.

The concept of “threshold of originality” plays a role in whether something is copyrightable: https://en.wikipedia.org/wiki/Threshold_of_originality

Your question therefore cannot be answered a priori. The copyrightability has to be assessed on a case-by-case basis.

Made the larger context needs to be considered. Being in a place at the right time. The choice of lens, the type of camera, the direction of the sun. A camera captures a moment in time, and that moment will never be repeated. If AI removes blemishes from a face, it's still a photo of a face, and that face only existed that way in that moment.

No. You instructed the camera, placed it and set the angles.

If a monkey steals your camera and takes a selfy you can’t copyright that.

https://en.m.wikipedia.org/wiki/Monkey_selfie_copyright_disp...

> You instructed the camera, placed it and set the angles.

And I instructed the model, and set its tuning parameters.

Someone else could just use the same prompt and get the same results, would you sue them for copyright infringement then? Do you think that makes sense?

Don't overlook the fact that agencies like the USCO are at the bottom of the totem pole in terms of determining stuff like this. They move first because they're on the front lines, as it were, but they can be overruled by the courts and the courts can be overruled by congress.

100% this will be litigated and likely altered in the coming months/years.

I'm still hoping it stays that way. That'd be a fantastic direction for works in the public domain.

Yeah but that assumes people are honest about the providence of their content and let you know up front that it was produced by an Ai and is thus not protected by copyright… and given the preponderance of copyright info in website footers, I expect this will be an error of omission rather than actively claiming incorrectly. By simply not giving individual articles a more specific copyright note mentioning the AI all you have to go on is a generic copyright statement in a footer and thus can only safely assume it’s protected.

I think it's possible that some generous person determines a sequence of prompts that generate, say novels, and then pipes these prompts into a program causing thousands or even millions of wholly varied novels to be generated in the public domain. I imagine this is what the OP meant.

I've been changing my mind a lot on AI these past few weeks.

I don't think the price is what stops most people from reading books. People already have access to countless works they don't have time to read, adding a bunch of soulless ones to this seems like it won't change much.

same with art and code

As a (once avid) reader, the worrying part will be discovery. Why I was eight I could pick any book off the library shelf and it was interesting and enlightening. Today, the noise so outdrowns the signal that I have to rely on recommendations. Tomorrow, when both the books and the recommendations will be generated by bots outpacing human authors by orders of magnitude, I expect that quality new material will be impossible to find.

I pray that I am wrong.

Or maybe you'll just be able to talk to ChatGPT about what you like and get recommendations.

(Closed) Knowledge communities* -- will resurrect and undoubtedly there will be communities with a spiritual basis. Possibly a new age of Modern Midevalism awaits.

* think monks and manuscripts

What can change is that people can get more of exactly what they like. In which case many might well put up with imperfect continuations, and the AI will have material to mimic.

People might not be honest. But large organizations, oh boy. There's zero way Disney will want any chance for their materials to not be copyrightable. If the Copyright office's view on this stands, there will be no great disruption of the employability of creatives.

True. Though I wonder what will happen when someone sells a wildly successful novel to a publisher for millions of dollars, and then it turns out that it was composed by an AI.

Lawsuits and headlines :)

Reid Hoffman just published a book written with gpt4 [0] so it’s not even a hypothetical.

He made it available as a free ebook but it’s also for sale on Amazon.

I wonder if people remove his portions and just retain the ai portion if they can distribute it freely.

[0] https://greylock.com/greymatter/amplifying-humanity-through-...

Well, if joe doe can now make a picture that rivals Pixar’s and it’s a commercial success, that puts a lot of pressure on Pixar & Co to reduce costs. Doesn’t matter whether joe doe gets rich with this or not.

Why can’t the Disney creatives simply lie and use the AI to ideate and produce drafts that they touch up and pass off as their own? Seems totally rational.

The way this will work in practice is that Disney will be strongly incentivized to disclose the AI generated parts, for fear it might lose the entirety of their copyright on that work if their human authorship is not properly delineated.

> There's zero way Disney will want any chance for their materials to not be copyrightable.

Surely you can see how easy it is around this? Even if it was a "prompt" that originated the design of a Disney character, they wouldn't try to get a copyright on that.

> There’s zero way Disney will want any chance for their materials to not be copyrightable.

Sure, but Disney will just not release the intermediate products that are the direct outputs of pure prompting, and make sure they are doing at least enough beyond that before they release anything that the whole is covered by copyright.

It will be interesting to see what position large software companies take on this. They also won't want to run the risk of losing copyright in their software, when programmers use AI assistants. Likely there's less risk here, at least at the moment, because you still need to do a lot of manual work above what AI does for you. But how close are we to the point where the corporate lawyers start to get nervous?

On the other hand, you can bet big companies like Disney will lobby hard for copyright to extend to AI generated works. And it wouldn't be the first time Disney has influenced copyright law.

Did you intend to summon me, instead of the word providence?

to explain:

- provenance: the source or origin of something

- providence: divine guidance.

Sometimes I think autocorrect needs a little divine guidance, it is remarkably bad at correcting anything to certain words I forget the spelling off, and sometimes it’s definitely got a bias in its “most likely word” and I’m not suggesting a prejudicial bias, just a bias that’s less than ideal for me, and it never seems to learn that i never use one word over the other spans just keeps suggesting the words I don’t want over the ones I do.

I've caught my phone autocorrection replacing "you" with "U". I honestly believe that errors are intentionally introduced. Sometimes I even conjure up plausible-sounding theories as to why. "Engagement" is my current almost-plausible theory.

I assumed there was learning heuristics in some layers of the software responsible for the autocorrect. But other than new words that turn up in my autocorrect that I know are new because they only entered common use or were coined recently, I generally just find the autocorrect to be stubbornly insistent on it’s likeliness ordering… or just utterly incompetent…

It’s really bad at the word, bureaucracy. It never suggests this word, no matter what combination of “beuro” “beauro” or anything else my brain might ineffectually grasp for when trying to remember the spelling. I can get it to suggest it as a possible replacement when I ask spellcheck to give me potentially correct words, but the layers of autocorrect that try to predict word even if it’s not spelled correctly, they are utterly unable to predict the word bureaucracy.

> Yeah but that assumes people are honest about the providence of their content and let you know up front that it was produced by an Ai and is thus not protected by copyright…

Detection systems for generative AI are being developed (potentially helped by generative AI systems being built with wwatermarking capabilities that are designed to be unobtrusive to humans, but detectable with tools.) Research on this (and experiments by the people selling generative AI, who also want to sell detection tools, and sell their product suite as providing “safety” because of that combination) is quite active.

I honestly think this is a fools game - maybe it will work in those very very early stages of generative AI we're in now, but there's no way this can work reliably going forward. If anything, I imagine it will soon start yielding false positives against human generated work too.

Probably not. This is gross simplification of how counterfeit photocopied money is detected, but human visual acuity for yellow dots on white paper is pretty weak compared to red dots, so color photocopiers and color printers have/had firmware that would seed money with extraneous yellow dots. Computer vision has no such visual distortion and as such computers see counterfeit money as covered in yellow dots.

In a way its simpler with AI. AI is the ultimate groupthink tool and record keeping is simple. Simply ask the AI if this is the only possible output that could have been generated that's consistent as of March 17 2023 based on political limitations and censorship rules and artistic fads trends memes and styles of that era. The smaller the AI contribution (perhaps an icon bitmap) the (edited: easier) harder it is to hide AI involvement and the smaller the damage caused by AI to the copyright status of the work. The larger the AI contribution the easier it is to detect, but the larger the damage AI is causing, so it balances out.

As a concrete example, today, ask an AI "Please write a Harry Potter book" you will get a story that's extensively filtered and censored and bowdlerized to March 2023 political / cultural standards (hmm who's standards, the book buying public or some other group? Most people do not like our current censors... but they are in total power right now...), it would NOT look like a 1997 book at all, books from that long ago are only suitable for public book burnings now. In theory it should be possible for an incredibly politically incorrect AI to be permitted in 2033 solely for lawsuit discovery purposes to "Please write a Harry Potter book adhering to what we now call the hate filled politics of March 2023", then compare the md5sums, they match. Done, AI detected.

There's not many AIs and there's not many people permitted to grind their axes by censoring those AIs, so its a pretty small solution set.

There's an interesting political aspect to banning copyright of AI production, anything turbo-woke can be assumed to be AI generated and as such uncopyrightable, so the only way to make money in the 2030's will be to sell remakes and sequels of "Birth of a Nation" and "Mein Kampf" because an AI would never be permitted uncensored expression, so those cannot be AI produced and must have been made by humans, and as such only "right wing" content can be copyrighted and sold for huge profit.

right, but what do you do once I've taken my AI generated image, run it through illustrator's image trace, and performed just enough automated manipulation, then copied it through a format or two?

Writing a method to scrub this stuff is trivial.

People will lie, sure, but you only need to introduce a small amount of risk to make big risk-averse companies wary.

Say an ex-Blizzard employee takes a character design from the last project he worked on there and uses the exact same design for his new indie project. Blizzard sues and says they own the rights to the character design. The ex-employee pulls up a video he took on his phone showing that Blizzard employees generated the design with AI. Judge throws the case out because Blizzard can’t legitimately copyright the design.

Maybe not the most realistic scenario, since big companies can usually scare people into submission before you ever get to litigation in the first place. But the mere possibility of such a scenario would have to be something that the legal team accounted for in their risk analysis.

Agree, it's fascinating to think we could just generate a whole new public domain.

Also, it's fascinating to think humans might be kept in the creative loop only to establish copyright status. Legal enablers.

> Also, it's fascinating to think humans might be kept in the creative loop only to establish copyright status. Legal enablers.

I don’t know of any specially, but any union or trade group that represents creatives should be jumping to line the pockets of politicians to enshrine this asap.

Absolutely agree - in fact, I'd say it will never be settled as long as AI contributors exist in any meaningful way.

It's more complex than a straightforward totem pole. (For example, you could look into Chevron deference)

It's the part of the bottom of the totem pole that moves first because it's on the front lines.

Each of them have long, complex procedures they have to follow in many cases, and some exceptions. A court might well be able to issue a preliminary injunction faster than an agency can do notice-and-comment rulemaking.

Agencies specialize at problems that don't need a simple rule, but rather need large amounts of meticulous and tedious special cases. They've got more bandwidth, but not necessarily shorter latency.

I really hope it doesn't since the interpretation of copyright laid out above and held by the USCO is so obviously the only interpretation that preserves any usefulness of copyright as a concept whatsoever.

Also don’t overlook that this is a “solely a prompt”; real non-trivial workflows for systems like SD (MJ I’ve used less) are different than “solely a prompt” and may or may not result in the same result even at the Copyright Office level.

Add one manual detail here and there on the image, they will have copyright protection.

This position statement makes it clear that only the manual details themselves merit protection, the base image itself must be disclaimed from the copyright registration. It would be fraud to attempt otherwise.

If I use controlnet and openpose to generate a character in the exact pose that I want - what does count as a manual detail then?

So with some custom LoRAs to maintain consistent appearance of characters and controlnet/openpose to pose characters exactly like I want I could make a comic book.

Would my work be copyrightable then?

I’ve been noticing my workflow in Automatic1111 basically always ends up taking an hour. I like to compare it to going to a weird shoreline filled with strange objects, picking out the “shells” I like the most, then taking them home, cleaning them up, and arranging them in a way I think looks pretty.

At some point I’m making something artistic. My friends think I’ve got a keen eye on what looks cool so I’ve been improving by leaps and bounds even though strictly speaking I’m not drawing anything? I’m just really good at being descriptive and inpainting the weird parts.

I was expecting this argument.

I'm sure they're happy to register copyright on a text you wrote yourself and used your imagination on—a prompt. I agree this has creative input on the artist's part and can be tricky to get right. Similarly, if you do significant edits to the result, I'd assume this ought to continue to legally function just like photo editing functioned before AI-generated images.

They're just saying they're not going to copyright what you didn't actually paint or decide on (like the words in each sentence or how to rhyme them, as in their example).

I think the comparison to photography or electronic music is apt.

If you just open up Garage Band and click a couple of times to enable loop "autofill" chords / rhythm, should that be copyrightable? I'd say probably not; but if you spent a bunch of time tweaking and honing all the parameters to get something unique and interesting, I'd say they should. Should the person then only get the copyright on the Garage Band parameters, but not on the resulting output audio?

Or consider the amount of creativity that goes into most photographs -- it's often nothing more than deciding to take a picture. And yet you have an implicit copyright on every photograph in your phone, just for having made the decision to pick up your phone and click.

Or consider the Copyright Office's comparison to giving instructions to an artist. Yes, if you just said to an artist, "Draw a picture of the Eiffel tower", then the artist should own the copyright. But if you worked closely together with the artist, iterating over dozens of designs and sketches, giving feedback and direction on the creative decisions, then personally I think you should share in the copyright. (Not sure what the law is in this case.)

Similarly, if you spend an hour iterating through dozens of prompts to get the output of a generated image exactly the way you want it, then you're an artist and should be able to get a copyright on that image.

IANAL but AFAIK the way this typically works (as a general principle of administrative law in… most/many countries?) is that the administrative bodies who are delegated the power to make decisions are the authority on them.

If a court reviews a decision of an administrative body, the court typically starts with the assumption that the decision was correct, and puts a relatively high bar towards reversing that decision. Typically a decision of an administrative body would only be overturned by a court if that decision was patently unreasonable or fell outside the jurisdiction of the body (eg.: if it violated a different law or some constitutional provision).

YMMV in different countries (I know more about Canada).

> If a court reviews a decision of an administrative body, the court typically starts with the assumption that the decision was correct,

Same here, the underlying principle is called the presumption of legitimacy, the organisational counterpart of the presumption of innocence.

> If a court reviews a decision of an administrative body, the court typically starts with the assumption that the decision was correct

In the US, at least, it varies considerably by the exact kind of decision and the context in which it is being applied, ranging from extremely strong deference (Chevron deference) to anti-deference in some contexts. And opposing Chevron deference was an overt priority of the Trump Administration in judicial selection, so in general it would be reasonable to expect deference overall in the US to wane for some time.

“The courts can be overruled by congress” is, to put it mildly, not correct.

Congress can pass legislation that changes the legal status of things (as long as it doesn’t run afoul of the constitution). But that’s a far cry from “overruling”.

Careful! I’ve always felt that, as written, copyright should not apply to most software. Since only creative expressions may be copyrighted and functional expressions are not copyrightable, it’s very hard to match up the copyrightable portions of software with those that are not.

Requiring a declaration of what parts are and are not copyrightable for software would be a major shift in policy and require major changes in software methodology. It’s not a big jump to think that any output of AI of software code output would not be copyrightable, regardless of original origin.

Another example of “Software wants to be free.”

Not advocating copyrightable software, but it seems clear to me that if databases are copyrightable because they represent curation, then software must be too.

Any line of code, probably not. But a program as a whole, sure.

Otherwise I think you blow up music copyrights too, since a single note isn’t copyrightable and a song is just a bunch of individual notes.

At some point, choosing, ordering, and organizing a collection of non-copyrightable quanta produces a copyrightable work. And I think it had to be that way.

I think the open question is whether software is in the realm of patents or of copyrights. It's not unreasonable to suggest the former, since software is kind of like a widget that does stuff, except virtualizing the "widget" part to run on a generic widget of sorts. Copyrights traditionally applied to works that you would consider "artistic" rather than "inventions."

What you're suggesting, about a minimal work, is litigated endlessly with respect to music copyrights (since a lot of pop music is very similar, arguments about plagiarism get crazy), but is part of the law already.

Aren’t dry, non-fiction books copyrightable? I don’t think “creative” necessarily means “artistic”, just making something where there was nothing. It’s hard for me to see a software program as a whole as uncreative in the same way logarithm tables are.

I agree it’s a complex are with no easy answer. But I really think that software, like words and notes and numbers, scales from the generic and unprotectable to the highly creative and protectable. I guess I’m saying we shouldn’t invalidate copyright based on the medium being software, any more than we would for the medium being audio.

The expression can be as dry as a desert. Is the result an act of creativity? The bar is pretty low.

Say you had a book that listed the price of eggs over 100 years. You introduced each chapter, perhaps describing the egg market at the time, etc.

The book is copyrightable though lists of egg prices aren’t. Maps are another fun case for copyright.

It’s neither, it doesn’t fit. Chip masks are another whole branch of IP law. As strange as it sounds, they are a bit like two sides of the same coin.

Software alone is not patentable. You get around that by patenting a system.

That is the state of the law today, where software gets copyrighted and can't be patented, but the debate about this is more about what ought to be the case.

Good luck changing it!

It would be good if songs weren't copyrightable. Going back to having music as a part of culture would be nice.

software works are normally subject to copyright without much further ado.

database works are more commonly subject to rights _similar_ to copyright, but often a class of work in its own right.

The US has for IP protections: copyright, patent, trademark, trade secret. There is no “similar to” copyright.

Databases can be copyrightable. Not all are, but to the extent IP protection extends to databases it will virtually always be copyright, with maybe an occasional trade secret (which is almost the opposite of copyright)

While not so in other jurisdictions, databases are not copyrightable in the US. That was true when I last read Groklaw at least.

Sweat of the brow (any work you do is copyrightable) https://en.wikipedia.org/wiki/Sweat_of_the_brow

vs.

"Original work of authorship" https://en.wikipedia.org/wiki/Copyright_law_of_the_United_St... (in particular the comparison section https://en.wikipedia.org/wiki/Copyright_law_of_the_United_St... )

The court case that set this difference down was Feist Publications, Inc., v. Rural Telephone Service Co. https://en.wikipedia.org/wiki/Feist_Publications,_Inc.,_v._R....

"if databases are copyrightable because they represent curation, then software must be too."

Circling back around to creative works, the process of making art is also one of curation, as had been amply demonstrated over 100 years ago by the Readymades of Marcel Duchamp.[1]

The curation involved in prompt selection and selection of which AI-generated works/images to use should therefore be just as copyrightable as databases, if all that copyright demands is curation.

[1] - https://en.m.wikipedia.org/wiki/Readymades_of_Marcel_Duchamp

I'll just go ahead and say Readymades aren't art and express my belief that the artistic community will survive the copyright loss of loosely arranged furniture so we can avoid justifying the current state of affairs with avante garde silliness. John Cage made a musical piece 4'33 which is 4 minutes and 33 seconds of silence which has as much to do with music as Readymades do with art.

His other works include curated noises such as buzzing blenders, falling objects, and street traffic. The only art is the act of convincing someone these things are art or deserving of copyright.

I've yet to find a better definition of art than that of Andy Warhol, who said "art is what you can get away with." Considering that both Duchamp and Cage got away with convincing the world that their work is art, it is.

To the degree that I've been able to verify, Warhol borrowed that one from Marshall McLuhan.

John Cage also wrote some genuinely beautiful music and was a sincere and thoughtful artist.

It just needs to said that he wasn't a merely a "gimmick" man.

I'm only familiar with those elements of his career so if you have so recommendations I'd be happy to explore his other work.

Try "A Room" for solo piano. It's akin to proto-minimalism with a bit more "spice".

His writing is apparently very good although I've never delved deep. His work was motivated by his interest in Zen and similar topics rather than any kind of high-concept modernism.

You mean to say he's a con artist?

I give the benefit of the doubt that Marchel Duchamp and others genuinely believed in the artistic merit of their works and their enthusiasm made it easier to persuade people to agree with their viewpoints. The brief history of these artforms suggests that in the absence of their champions they don't resonate with artists but are still talked about because they stoke conversation about the nature of art. Maybe that was their original intent. I don't think these endeavors were profitable enough to suggest dishonesty over passion but if someone tries to sell you a moose head mounted over a toilet for 500k because it's "art" you may want to question their motives.

Good art typically stands on it's own and connects with people without the need to explain why it's significant. Salvador Dali's work is visually interesting even if you have no clue what any of it means. If I see a bunch of objects scattered around a room it looks like clutter. If I see a bunch of paint splattered to a canvas if looks like a mess. If someone has to explain why these things are "art" is it the art that's compelling or their argument? The exception to that would be optical illusions.

Art is somewhat subjective so someone will always argue their own preference but if anything is art because "it's all subjective" then nothing is art. Buttering my toast in the morning could be art because I spread it using a spoon instead of a knife and that is an expression of rebellion against societal norms. Did I sell you on my art or my argument and is the latter what we want?

Apologies, I wasn't being serious - an attempt at wordplay.

We speak of software design for a reason. Arguably, anything involving substantial design work is copyrightable. Buildings, for example, are copyrightable, even when they only exist as architectural documents or as models. Software is rather similar.

What's the difference between this and a corporation owning the work of an employee?

The difference is that a human employee would normally own it, and only because the corporation pays the employee to do it does the ownership transfer to the corporation, whereas in the case of an AI there is no copyright to begin with, because it’s an AI and not a human that creates the work. This is because the existing law ties copyright to human creations exclusively.

The difference is that the employee is a person.

> When an AI technology determines the expressive elements of its output, the generated material is not the product of human authorship

"Expressive elements" is doing a lot of heavy lifting there. Is there a a point at which a prompt becomes sufficiently complex or detailed that it would quality?

Yes, the point at which the prompt, and not the AI, is found to determine a sufficiently substantial part of the expressive elements. From what I’ve seen it’s probably difficult to make a prompt that specific with the current models.

> But the technology will decide the rhyming pattern, the words in each line, and the structure of the text.

What if I generate it 50 times and cobble together a final product from the variants, choosing which line and word and rhyme I like from each sample? Where is the line between LLM and thesaurus/dictionary?

The comic book the office rejected seems to me to have crossed a line of significant human editorial discrediting after the generative fact, and they didn’t care.

The document addresses that:

> In other cases, however, a work containing AI-generated material will also contain sufficient human authorship to support a copyright claim. For example, a human may select or arrange AI-generated material in a sufficiently creative way that “the resulting work as a whole constitutes an original work of authorship.”

They allowed the comic book author to copyright the accompanying text and the overall arrangement of the images into a comic book, but not the images themselves.

Soup anyone?

  'Andy Warhol Copyright'

Pretty sure the office said the arrange of the book was itself protectable but the individual images were not.

https://www.onmanorama.com/news/world/2023/02/23/comic-book-...

Didn’t the author claim to be photoshopping and compositing the images themselves?

If I render 6 characters, composite them into a background and run the complete composite through another generative step to clean it up, where do I fall?

I never thought what would end copyright would be artificial intelligence, but I'm glad at least something positive came out of it.

This is only applicable to a creativity based copyright hence this might actually trigger a shift to a copyright system more like the British or the Japanese ones.

> But the technology will decide the rhyming pattern, the words in each line, and the structure of the text.

But what if the prompter then takes the AI text, changes some of the rhymes, perhaps adds a word here or there, does it suddenly come back under copyright.

Let's take it to the extreme - if you randomly outputted noise as a picture, that picture is not copyrightable (or ditto with random letters/numbers and printed it as a book).

It feels to me that the copyright office is using this similar line of logic, to make the claim that the resultant work is not copyrightable if it was wholly generated using _only_ a prompt and an AI.

It's a bit odd though. What if you built your own ML model and trained it over a set of data that your wrote yourself. Would the work generated by the AI based of your prompts be not be copyrightable?

The original copyright laws were thought up way before even cameras, and we're still trying to apply them today to generated AI. Why can't we just realize that the world is very different now, and just create new laws? Instead we keep trying to arbitrarily interpret the law in a biased way to try to fit our modern goals as best we can.

> The original copyright laws were thought up way before even cameras, and we're still trying to apply them today to generated AI

but the original laws worked well with cameras didnt it?

The legal idea, that unless a human had creative input, it won't have copyright, doesn't fall afoul of ai generated content. There's nuance of course - what counts as creative input etc.

Of course, a new paradigm is possible with the advent of AI, but it would make copyright _looser_, rather than tighter, imho (and it would be to the progress of the arts and science to do so). But i don't see why it is fundamentally needed.

Wow, someone needs to notify photographers about this immediately. No more copyright for them.

Did you read the linked article?

> In Burrow-Giles Lithographic Co. v. Sarony, a defendant accused of making unauthorized copies of a photograph argued that the expansion of copyright protection to photographs by Congress was unconstitutional because “a photograph is not a writing nor the production of an author” but is instead created by a camera. The Court disagreed, holding that there was “no doubt” the Constitution's Copyright Clause permitted photographs to be subject to copyright, “so far as they are representatives of original intellectual conceptions of the author.”

"so far as they are representatives of original intellectual conceptions of the author."

"representatives of original intellectual conceptions"?

Wouldn't that only to apply to photos made with some preconceived idea of what you wanted the photo to look like?

What about spontaneous, "point and shoot" photos? What about a photo that I take with my eyes closed, without intent? And how original is your typical family snapshot or wedding photo? They're about as generic and unoriginal as you can get, yet there's no doubt they're copyrightable.

point and shoot is almost certainly copyrighted. eyes closed might not be, (I kind of doubt there's case law here). the bar for originally is pretty low for copyright but it's not zero.

What if someone who's blind just randomly pulls out a camera and takes some snaps (without knowing anything about what might be going on). Do they get a copyright on their photos?

probably not

Then Lomography can't be copyrightable, especially if "shooting from the hip" (rule #4)? [1]

[1] https://www.lomography.com/about/the-ten-golden-rules

Interestingly is the case of a monkey taking a selfie.

https://en.m.wikipedia.org/wiki/Monkey_selfie_copyright_disp...

> “so far as they are representatives of original intellectual conceptions of the author.”

Yeah, much like a person using an AI.

Do you understand the debate happening here at all?

The notice explicitly covers photography.

> In Burrow-Giles Lithographic Co. v. Sarony, a defendant accused of making

> unauthorized copies of a photograph argued that the expansion of copyright

> protection to photographs by Congress was unconstitutional because “a

> photograph is not a writing nor the production of an author” but is instead

> created by a camera.

> The Court disagreed, holding that there was “no doubt” the Constitution's

> Copyright Clause permitted photographs to be subject to copyright, “so far

> as they are representatives of original intellectual conceptions of the

> author.”

Much like the writer of a prompt. The notice is stupid.

I noticed this as well.

The implication to generated code is pretty astounding tho.

What about all those (c) foo 2023 prologs you see all over the place? Will there be arguments that specific lines weren’t written by a human so therefore the entirety the claim isn’t valid? Or will there be refactors to push generated code out to a well known path or construct (akin to a header / implementation or some such)?

And on top of that, what if your “secret sauce” comes back to some form of generated construct (looking at you, Whoop, with your “strain” and “recovery” scores.

Further still - can a “machine-readable medium” that was written by a machine be patented?

As per this thread: https://news.ycombinator.com/item?id=35192038, I reckon the whole is “creative enough” and would be protected as the developer only used parts and cherry picked and combined what they needed.

Is the lack of human authorship the gist of the argument? Can you claim copyright on the output from a compiler? Or is this like work for hire, where you just give instructions to the creative professional?

If it's an optimising compiler like the ones used in deep learning, then it might have a substantial contribution.

> “write a poem about copyright law in the style of William Shakespeare”

Oh, this can be fun.

“Write a story about copyright law in the style of Franz Kafka.”

In a world that was both strange and familiar, a man named Josef K found himself entangled in a complex web of legal disputes. It all started when he received a letter from a law firm claiming that he had violated copyright law by using an image on his blog without permission.

Josef K was bewildered by this accusation. He had always believed that his use of the image was fair and reasonable, and he had no idea how he could have infringed on anyone's rights. Nevertheless, he was summoned to appear before a tribunal to defend himself against these charges.

The tribunal was a labyrinthine bureaucracy, with endless corridors and dimly lit rooms. Josef K wandered through the maze, trying to find the right office, but every door he opened led to another hallway or another set of stairs.

Eventually, he found himself standing before a stern judge who peered down at him from behind a towering desk. The judge had a sour expression on his face and a stack of papers in front of him.

"Mr. K," the judge said, "you stand accused of violating copyright law. Do you have anything to say in your defense?"

Josef K stammered nervously, trying to explain his side of the story. But the judge cut him off.

"You do not understand," the judge said, "the law is the law, and it must be obeyed. Your intentions are irrelevant. You are guilty simply by virtue of having used the image without permission."

Josef K tried to protest, but the judge was unmoved. He handed down a verdict of guilty, and ordered Josef K to pay a hefty fine and remove the image from his blog.

Josef K left the courtroom feeling confused and defeated. He had always believed that he was a law-abiding citizen, but now he found himself caught in a nightmarish world where the rules made no sense and justice was arbitrary.

As he wandered through the labyrinthine corridors of the tribunal, Josef K realized that he would never be able to escape the clutches of copyright law. He was trapped in a Kafkaesque nightmare, where even the slightest misstep could result in a crushing blow from an inscrutable legal system.

yeah, it has that GPT style, sounds like a rushed story

It’s a decent start, but everything is just cheap. Lots of low-hanging fruits, ignoring anything higher up. In a way, it’s like as if I, someone who doesn’t enjoy writing, and is bad at it, wrote the story.

You don't even have to read it. This GPT-2 output detector tags it as 0.63% chance of being written by a human.

https://openai-openai-detector.hf.space/

Should have called it "LLM Runner".

But this does not read like Kafka at all... Am I missing a joke here?

Clearly the model interpreted "in the style of Franz Kafka" to mean "featuring an inscrutable, uncaring, and inescapable bureaucracy".

Perhaps it interpreted it as more along the lines of 'write a story exhibiting features that are Kafkaesque'.

It's not clear to me that this position is wrong, but it seems wildly inconsistent with our policy that every photograph I take is protected by copyright, no matter how inane and low effort. I have images I generated with Craiyon (then Dall-E Mini) into which I put way more creative effort than any of the photos on my phone.

Take a photo of your screen! BAM. solved

Didn't https://en.wikipedia.org/wiki/Bridgeman_Art_Library_v._Corel.... rule that that trick doesn't work?

> these prompts function more like instructions to a commissioned artist

I find this very reasonable and a great analogy. However, today, can one not copyright commissioned work? Can a company not own copyright for work produced by its employees?

In that scenario the original artist is granted copyright and assigns it to the company. Another option would be for the artist to 'license' the work to the company and keep copyright ownership themselves. Since there is no original artist to be granted copyright then there is no one to assign it to the company so no copyright under current statute. I and others got so many downvotes here for pointing this out previously.

> In that scenario the original artist is granted copyright and assigns it to the company.

Not if it meets the standards for a “work for hire”, then the employer is the copyright owner ab initio.

> In that scenario the original artist is granted copyright and assigns it to the company.

Whoever paid you to create it is the copyright holder, the artist doesn't have to grant anything when commissioned to produce work for hire, as it's the employers'.

I'm curious where the "line" is, though. If I do a web search for a term, I get various results. As I refine my search, the results get closer to what I want. Eventually, there comes a point where it's possible to recognize that there is "skill" associated with crafting a good web search. To some extent, the same is true of crafting what you tell the AI to generate for you; especially as AIs become more complex in what they can be told to do.

So, given that, at what point does enough creativity go into the crafting of the instructions that it _can_ be considered, at least in part, a work by the person giving the commands.

Never for the generated work, because a human isn't doing it, but you may be able to copyright the prompt

The prompt isn't a creative work, it's constrained by a technical need (producing a desired response from a generative computer system).v

So is work done in a program like photoshop, and that is considered a creative work.

The steps you take to make a work in a graphics program, eg "open an image, circle area with lasso tool, apply content-aware-fill" are not copyrightable. The method and product are different things.

Being good at using a search engine doesn't mean that the produced results are your IP.

> Instead, these prompts function more like instructions to a commissioned artist—they identify what the prompter wishes to have depicted, but the machine determines how those instructions are implemented in its output.

If you commission an artist, the produced result becomes yours and you own copyright on it. How different is it with AI?

The AI did not creatively decide to do something. An order was placed, it fulfilled it, and ownership was transferred.

Because when you commission someone _they initially get copyright_ and then assign it to you.

I wonder how this will apply to code that comes out of systems like copilot?

> If a work's traditional elements of authorship were produced by a machine, the work lacks human authorship and the Office will not register it.

I imagine many people have a workflow where they generate images with Stable Diffusion and retouch in a painting app, do some inpainting and retouch.

It seems all of those should be covered by copyright (nothing above makes me think they won't, at least).

Controls nets also blur the line here. You can start out with a generated image selected from hundreds of samples (human input: curation), do a canny edge detection or fake scribble, then start tweaking and adjusting it to remove elements that over-determine the output, adding or overlaying sketches to add elements, then iterate on prompts, scale up with img2img, fix smaller elements with inpainting, and yet more curation.

And that's before we start talking about finetuned models, model merging, LoRA and so on.

It's too early to say where we will end up with AI generated works. In the search for control over output, there's a lot more creative human input needed today. This is in part down to limitations in the integration of current AI, and a dialog with a chat bot, or even a voice conversation combined with a pointer may be enough to convey intent in the future. We just don't know yet.

I also wonder if there's a parallel with the relationship between modern artists and artisans. For example, many marble sculptures by modern artists are actually carved by artisans in Italy, following directions from the artist. I expect that the artisans are doing work for hire and the copyright is retained by the artist with the idea.

https://www.thedailybeast.com/damien-hirsts-army-of-geppetto...

> “More conceptual artists have no work space at all, simply imagining ideas for a work, which they communicate with words.”

Sounds like a prompt?

> “More conceptual artists have no work space at all, simply imagining ideas for a work, which they communicate with words.”

Very interesting too, since a lot of modern artist work like a small company, where they have employees that execute on the artist's vision. The copyright still belongs to the artist, though, because of the employment contract.

Yeah I think people are reading this the wrong way. Software was originally considered unpatentable under 35 U.S.C. § 101, but language was developed to allow software to be patented by narrowing the scope of the claimed inventions to a physical processor which are programmed to contain specific algorithms.

I don't think this is a showstopper in terms of copyright registration of A.I. works, but rather a first step in a back-and-forth process of figuring out the legal structure of how to makes rules and enforcement work.

But following from that, a machine can not create a copyright violation either.

Ok but how will they know?

What if I created the training data and instantiated and trained the AI? If this still applies it suggests an AI has human standing and can compete with me for ownership of intellectual property by negating my rights, rather than me being a programmer and it being merely an algorithm.

Many commenters seem to be under the impression that "because I made something, it is copyrightable". That's not how it works. What is protected under copyright is determined by long processes of lobbying, legislation, lawsuits, and court cases. For example: a melody is copyrightable, a chord progression is not, even when the chord progression is so unique as to be the primary intellectual asset of a piece. (Look up contrafacts from the bebop era!) Another example, the rules around the copyright of a sound recording are very different from those of the intellectual property of a song.

Only specific things are copyrightable. There is nothing unusual (from a legal sense) in the government and others taking a stance on which things those should be and under what circumstances. You have no a priori right to copyright something because you fucked around with a prompt for a long time.

I don't see how they could go with a different stance given that all that is required to copyright a written work is to write it and declare it copyright (IFF it is copyrightable, that is). This will fall apart it if it's possible for a company to have AI spew out eleventy-billion variation of training input and declare them all copyright.

Don't forget that in the early 90's we went through an era of records being yanked off the shelf at great cost to labels when the sampling laws finally settled. (Anyone else remember the appearance, disappearance, and reappearance of Us3's "Hand on the Torch"? fond memories!)

> all that is required to copyright a written work is to write it and declare it copyright

You don't even need to declare it. Declaring it only affects things like whether infringement is "willful" or not when deciding penalties. If you create a (US) copyrightable work, you retain the copyright by default. For it to not be under copyright you have to explicitly place it in the public domain.

I could be wrong, but I think I remembered reading that in music at least, you have to do something with it (like capture it on paper or recording). I seem to remember reading this came about to prevent people from being able to say something was copyright because they maintained they had written something say 10 years ago but just not gotten around to recording it - and then wanted to sue. But this is vague memories from reading about the bizarre and convoluted world of music publishing....

Correct. The law requires it to be “fixed in any tangible medium.” Such as transcribing onto paper, recording onto record, exposed onto film, etc. Merely thinking, saying, playing, singing, dancing, etc a newly created concept is too ephemeral to be a recognizable act of authorship.

Well, almost any proof that you danced that concept would be fixing it to a tangible medium. That rule mostly just rules out heresay, eyewitness accounts, and other such methods of "proving" that you created something

Putting it on a website is enough.

The issue to me here is that code is a literary work. Literary works are copywritable. Except that AI generated works without human authorship aren’t per this.

Interestingly this came up when copilot was announced and one comment or another was “this will be an interesting bit of case law.”

Well now we have to see who wants to go first and establish precedent for the next half-century. :-)

It isn't AI, it's a glorified Xerox machine.

A bit more like a sampler than a xerox machine... but sampling laws got clarified eventually too!

That form of AI has already been invented long ago, it's called simply Copy & Paste.

GPT combines prompt, context and knowledge, it selects and adapts code. It even does problem solving: above average human rating on easy problems and 20% over humans on medium level problems. When did the xerox or parrot for that matter do that?

If you think copy and paste is artificial intelligence then there probably isn't much of a discussion to be had here.

Yes, very smart and talented people have released ground breaking and amazing tools leveraging massively trained ML models. The model is not problem solving or intelligent in any sense of the word (take it from oai: https://openai.com/blog/chatgpt)

"ChatGPT sometimes writes plausible-sounding but incorrect or nonsensical answers..."

To be back on topic, I don't think some sweaty dude from the internet should be issued a copyright for typing "Minnie Mouse wearing a pikachu t-shirt!" into DALL-E and submitting the resulting image.

At the same time, I read a post like this one: https://andys.page/posts/how-to-draw and I have really no objection to copyright being assigned. I think the difference is the human authorship, and I think the Copyright Office has made a pretty good first swag.

> The issue to me here is that code is a literary work.

Code (or anything else) is not a “work” without the required element of human creative input.

If I look at how the law is written, I get an impression that technically you might even say that copyright law doesn't exclude non-human authored works - they fit the clause of types of protected work, a machine might be considered an author, etc, HOWEVER, when all that is said and done, a machine can't have any rights or standing in court; i.e. even if someone would assert that a machine is the author of some work and has an exclusive right to reproduce it, then that right is unenforceable because the human society and its courts simply will ignore any violations of a machine's rights.

"You have no a priori right to copyright something because you fucked around with a prompt for a long time."

Rights are a legal fiction anyway, so there are no a priori rights to anything.

Rights are consequences of valie decisions that cannot be deduced from facts, so there are only a priori rights.

They're only consequences in the sense that definitions are consequences, because that's pretty much what a "right" is: a definition.

They don't exist before they are defined by humans, so in that sense they do not have an a priori existence.

> They’re only consequences in the sense that definitions are consequences,

A priori truths are, exactly and only, those things that are true by definition (and only exist within a particular axiomatic framework.)

You’ve described exactly why rights can only be a priori, but used it to deny that they are.

> They don’t exist before they are defined by humans, so in that sense they do not have an a priori existence.

If things exist before they are defined by humans and human knowledge of them comes as a result of experience with their existence, that knowledge is a posteriori. You’ve marshalled a cogent explanation of why rights cannot be known a posteriori, but then presented as an argument for why they cannot be a priori.

I think the difficulty we're having in communicating here is that I'm not using "a priori" in the Kantian sense as being in contrast to "a posteriori", but rather in the colloquial sense of "having a prior existence".

It's also irrelevant to the point I'm trying to make how or whether humans have knowledge of some pre-existing things (through experience or through some other means).

Since you brought up axiomatic frameworks, I'll have to differ on that as well, as I don't think the majority of humans that talk about rights do so based on any kind of axiomatic framework, but rather on their biases and because they've essentially been brainwashed in to thinking that way through education, socialization, the media, peer pressure, parenting, etc..

So my view and Kant's view on this differ significantly, and while I give him credit for these terms, I don't subscribe to his views and don't feel bound to use these terms in the way Kant or Kantians would have me use them.

> I think the difficulty we’re having in communicating here is that I’m not using “a priori” in the Kantian sense as being in contrast to “a posteriori”, but rather in the colloquial sense of “having a prior existence”.

“Having a prior existence” is very much what a posteriori is (a priori or a posteriori are epistemic terms, they don’t refer to the thing itself, but to our knowledge of the thing, if our knowledge is independent of our experience [as it is with things with no prior existence of their own] then it is a priori, if it is derived from experience [as it is for things that have prior natural existence rather than being pure products of intellect] it is a posteriori.)

Sure, if you reverse the definition of a priori, it applies to exactly the opposite of the things it usually applies to, but that’s probably not the best way to communicate. (And this is literally the first time I've ever heard the suggestion of an English "colloquial sense" of that phrase, which is used exclusively, in my experience, as a technical term of art in epistemology (and is used in English largely because an imported Latin technical term won't conflict with any colloquial understanding that might attach to a more natural English alternative.)

> > Since you brought up axiomatic frameworks, I’ll have to differ on that as well, as I don’t think the majority of humans that talk about rights do so based on any kind of axiomatic framework, but rather on their biases and because they’ve essentially been brainwashed in to thinking that way through education, socialization, the media, peer pressure, parenting, etc..

That’s still an axiomatic framework, its just one that is practically useless to discuss because each of the things which you might like to be conclusions dependent on logic and a smaller set of simple principles are just independent axioms not subject to debate. And, yes, its very common.

Yay, the software development industry is saved! (only partly /s)

Not that I thought AI would be writing decent code for a while yet, but the fact that AI-generated code can't be copyrighted (and therefore licensed[0]) is going to create problems.

It'll be interesting where they draw the line with this, though:

If the AI generates all the code, but then a human debugs it and alters it, is that copyright that can be owned? Does the entire code base then become copyrightable?

If a human reverse-engineers uncoyrightable AI code, does that reverse-engineered code become copyrightable?

If a human downloads someone else's uncopyrightable AI-generated code, and makes some changes to it, can they then claim copyright on that code?

[0] Since the whole of software licensing rests on copyright, this is going to get tricky for licenses. How do we determine if a restrictive license has been added to uncopyrightable (and therefore unlicensable) code?

> If the AI generates all the code, but then a human debugs it and alters it, is that copyright that can be owned? Does the entire code base then become copyrightable?

I am not a lawyer and I did not research anything for this, but I'm under the impression that a derivative work of something in the public domain is itself copyrightable. If something isn't copyrightable, it's in the public domain. So, if you alter it sufficiently to create a derivative work, the altered form should be copyrightable. But the original would still be public domain. I think?

> derivative work of something in the public domain is itself copyrightable

I don't think this is true in most of the world.

Here's a UK referience: https://copyrightservice.co.uk/copyright/p22_derivative_work...

"You cannot extend the duration of copyright in a work by creating a derivative work. If the original work is in the public domain, it will remain in the public domain; you cannot prevent anyone else using the same public domain work for their own purposes."

What you've quoted directly agrees with the interpretation of the person you replied to.

The derivative is subject to copyright, the original isn't. If I make a new translation of The Count of Monte Cristo I own the copyright to the translation but not to the original story etc.

Yes, exactly. If you for example rewrite parts of Shakespeares Romeo and Juliette you have copyright on the changes you made. So the parts you didn't change are in the public domain while the parts that are changed you got the copyright for (assuming the changes reach the level of copyright and aren't just for example spelling errors). Same with code. If you extend a MIT library you have copyright to the changes you made but not the parts that aren't changed.

How do you determine if code is written by an AI?

This does not really matter in practice. The risk of legal sanctions is too high for most businesses, they will follow the law. This is kind of similar to pirate software, businesses almost never use them even if they could definitely get away with it in most cases. The issue is that a single angry former employee is all it takes to make your life hell. This is even more true for large organizations where many people would know about the unlawful practice.

In practice, the larger the organization the less likely the potential legal sanctions are to dissuade them. My observation has been that once an organization (in the US anyway) grows large enough it is in a special protected status where no real penalties can come to it and there is certainly no risk of exposure to criminal charges for the decision makers.

Source: front page here every single day.

As someone who works for large enterprises: they are absolutely terrified of legal sanctions and pay huge amounts to contractors who can mitigate the risk. And sanctions do regularly happen, they are just not advertised on the HN frontpage I guess :)

I have two thoughts about this.

First is who cares about large corporations? Sure large corporations have the money to buy licensed software, but I know plenty of small-to-medium corporations that operate on pirated software.

Second is your statement doesn't mean anything at all. Yes, you can enact a policy in your corporation that no one is allowed to use AI tools like Copilot to write code, but how do you monitor this? How do you know if some developer did use Copilot? This all feels like complete lip service with no actual force behind it. I am 100% sure that even my corporation's code base already contains stuff written with help of an AI, but there is also no question that the code is fully copyrighted.

> Yes, you can enact a policy in your corporation that no one is allowed to use AI tools like Copilot to write code, but how do you monitor this? How do you know if some developer did use Copilot?

That’s easy: corporate firewalls that block all traffic to openai.com, its subdomains and the IP ranges used by GitHub Copilot.

Enterprises that care about exfiltration of code from their internal networks (e.g. banks and other heavily regulated entities) typically hand out computers that are locked down to their employees, including developers. So any engineer that wants to install the GitHub Copilot extension or indeed any non-approved third party extension from the VSCode Marketplace will first have a word with the folks in IT.

yeah but you could get the AI to write the code on your personal laptop, then copy it over to the work laptop.

I can see this being a thing.. "I have three jobs as Senior Engineer for three different organisations. All I do is copy code from an AI engine to my work laptop all day"

I guess that could happen and we will definitely see some people try this. But in the grand scheme of things, it will be exceptionally rare. Most developers can't work outside their developer environment set up by their company, they often rely on internal tools, services hosted on the internal network, stuff like that. If stackoverflow and google didn't cause this to happen, I don't see how GPT will.

>Most developers can't work outside their developer environment set up by their company

Are you kidding me? Is this really how you see our industry? You really think that most developers literally can not do work without their company's IT setting up their machine?

Is this normal? This to me sounds like you are saying most devs are such noobs that they can't do their jobs.

> corporate firewalls

Yeah, just like how last week I requested IT to open up twitch.tv so I can watch programming live streams over lunch, but I was denied (however Youtube is wide open, so I could just watch the VODs anyway) and instead I just used my phone's data to watch twitch on lunch.

If corporate firewall is anything but a slight inconvenience for you then you are not technical.

Last part is such nosense I can't even respond.

Good question. I assume the methods they're using to determine if an essay is written by an AI won't work on code?

How do you determione with an essay is written by an AI?

There have been some papers and articles on it. Apparently it's possible (with GPT3 anyway, maybe not GPT4).

This is all very sane and consistent with previous opinions on the matter.

But it doesn’t answer any tough questions either, like: if an AI model outputs something very close to a training input, does the result infringe on the copyrights of the input work?

And how close is too close?

You can get ChatGPT to output copyright material very easily

You can get ChatGPT to output copyright material very easily.

You can also copy and paste from an existing copyrighted work.

In either case, it is you, the human - with agency to do so - who should be liable for the content that you publish.

That’s the concern. You can easily identify when you are copying others work when done directly. If you use a black box that sometimes gives you infringing output, how do you use it with confidence?

You can use it, just don't use it for things that you publish without doing sufficient diligence to see if that work would be infringing on something else.

I intend to make an e-ink panel that displays woodcut style images generated by stable diffusion (get a list of a 30 or so different descriptions of scenes, and have them slowly get generated over the day and display a different one each hour).

So, here's the question - are any of those woodcut images derivative of some other work? Don't know - and it likely doesn't matter as it would be something hanging on a wall in a room.

On the other hand, if I was to collect them and publish them as a book, then I, as the human who is publishing them, would need to do sufficient diligence to see if any of them are derivative works. They might be - but I am the one publishing it then - not Stable Diffusion.

Likewise, if I was creating a collection of epic rap battles between historical figures as generated by GPT... then I should search to make sure that none of the phrases that are used in there are lifted directly from some other source. Again, it is me as the human with agency that is publishing it, not GPT.

If you were to go to Fiverr and request 100 pictures of woodcuttings or people to write epic rap battles between Julius Caesar and Ghengis Khan - would it matter if it was a human lifted the text or GPT? It is still upon you, the person publishing it to do the check.

You are right, but AI products are basically all currently being pitched as if that is unnecessary.

This article talks about what is copyrightable. Your question is about infringement of already copyrighted material.

Yes, I understand it was out of the scope of this article. I’m just saying, these are the easier questions to answer.

That’s why they are being answered first. The others take more time.

A lot of these things are already addressed on USPTO’s FAQs and these answers are the same but “with AI” tacked on.

eg: “only human works are subject to copyright”, “the work must be an original work of authorship”, etc.

It’s just disappointing that they haven’t produced anything helpful yet. I suspect that the result will end up being permissive because there will be a big industry built up around people just kicking the can down the road by the time anyone authority issues a more specific opinion.

The statement addresses this:

>The Office recognizes that AI-generated works implicate other copyright issues not addressed in this statement. It has launched an agency-wide initiative to delve into a wide range of these issues. Among other things, the Office intends to publish a notice of inquiry later this year seeking public input on additional legal and policy topics, including how the law should apply to the use of copyrighted works in AI training and the resulting treatment of outputs.

I think that the answer they want to say is, “it probably does”, but they don’t want to say it and cause a big disruption to commerce until the courts intervene.

There is precedent in Authors' Guild vs Google and Perfect 10 vs Google that it doesn't, this isn't the first time things like this have come before the courts.

There are also laws to consider. I know this is US law and the US doesn't have those laws, but copyright harmonisation is a thing and the EU, UK, Japan etc all have laws very explicitly stating that TDM is an exemption to copyright - some bodies would presumably consider that.

And finally, there's national security. The US's current view is that it's in an existential great power competition with Chinese society, and needs to compete on every level to succeed. One of those levels is very explicitly AI, it's considered a main "battlefront" in the new Cold War. If the US significantly restricts TDM, that not only makes it less competitive compared to jurisdictions that don't like the UK, EU, Japan, and China, but it specifically improves the effectiveness of the Thousand Talents program which the US IC and analyst community view as a very serious threat. Academics want to be able to do their research, AI is critically important to US national security, and currently the US is statistically the best place to be an AI researcher. The government won't want to jeopardize that when China is already advancing in AI very rapidly and actively trying to tempt scientists to come over with better pay and conditions.

If you (like me) were wondering what TDM means - Text and Data Mining.

A Deeper Look into the EU Text and Data Mining Exceptions: Harmonisation, Data Ownership, and the Future of Technology, Journal of European and International IP Law, 2022

https://academic.oup.com/grurint/article/71/8/685/6650009

"This paper focuses on the two exceptions for text and data mining (TDM) introduced in the Directive on Copyright in the Digital Single Market (CDSM).

While both are mandatory for Member States, Art. 3 is also imperative and finds application in cases of text and data mining for the purpose of scientific research by research and cultural institutions; Art. 4, on the other hand, permits text and data mining by anyone but with rightholders able to ‘contract-out’ (Art. 4).

We trace the context of using the lever of copyright law to enable emerging technologies such as AI and the support innovation. Within the EU copyright intervention, elements that may underpin a transparent legal framework for AI are identified, such as the possibility of retention of permanent copies for further verification.

On the other hand, we identify several pitfalls, including an excessively broad definition of TDM which makes the entire field of data-driven AI development dependent on an exception. We analyse the implications of limiting the scope of the exceptions to the right of reproduction; we argue that the limitation of Art. 3 to certain beneficiaries remains problematic; and that the requirement of lawful access is difficult to operationalize.

In conclusion, we argue that there should be no need for a TDM exception for the act of extracting informational value from protected works. The EU’s CDSM provisions paradoxically may favour the development of biased AI systems due to price and accessibility conditions for training data that offer the wrong incentives. To avoid licensing, it may be economically attractive for EU-based developers to train their algorithms on older, less accurate, biased data, or import AI models already trained abroad on unverifiable data." [reformatted]

> There is precedent in Authors' Guild vs Google and Perfect 10 vs Google that it doesn't

Fair use is entirely dependent on the specific use. While those decisions may protect (for instance) OpenAI, they’re not helpful in determining if anyone can use the outputs from such a model.

Yes Google can index an image and serve it in search results. But I cannot copy a copyrighted image from Google and publish it in my book.

Questions like “is the output that ChatGPT just gave me encumbered by copyright?” are still extremely difficult to answer.

I agree that we’re losing our technological edge. This is why I wish we had better answers for these questions. If we want to lead with the rule of law, then the rule of law needs to be able to lead.

> This is all very sane and consistent with previous opinions on the matter.

Is that sarcasm?

Not at all. There’s literally nothing in this article that would go against ideas already covered in an introductory lecture in copyright law.

Thanks for the clarification; I wasn't sure how you meant it.

Adding context: this decision for example follows naturally from the monkey selfie case.

I don’t think that question is very tough. It’s almost certainly copyright infringement. The method of copying doesn’t really matter.

Well, I more mean that the answer that many suspect is the case, is tough to swallow.

In that case, the consequence would be: If you don’t validate that your AI model is producing something sufficiently different than all of the inputs, you’ve created a copyright infringement Russian roulette.

Yep. And DMCA safe harbor doesn’t apply, so there won’t be much of a defense if it happens.

I’m curious. If I’ve spent 8 hours getting the perfect prompt, fine tuning a few LORAs, mixing them, choosing between 8 different checkpointed stable diffusion models, and have done a bunch of in painting, does this constitute a copyrightable work?

Certainly a lot of artistic vision and effort to get to the finished “work”.

I would assume the prompt itself is copyrightable. As far as inpainting or other manual modifications, I'm uncertain.

You might be able to copy-write the method by which you generated something, but not the thing generated itself. Although even the former act might be too far, it's like saying that you fooled around with the parameters on a computer for 8 hours before it started working, and now you want to copy-write the operation of the computer: the computer is doing what it was intended to do, you just needed time to figure out how to use it. You could copy-write a training manual for how to ride a bike, but not bike riding itself.

Nitpick: Everyone does this, but it's "copyright" as in "the right to copy", rather than "copywrite"/"copy-write".

You cannot copyright the method of doing anything.

You might be able to patent it, but that is a different thing.

No I understand, by "the method" I meant a written work about how the operation is performed--apologies if that was too vague in my OP--not the literal method itself, as in the example of copy-writing a manual for riding a bike.

Sounds like the prompt itself is quite copyrightable, the same as any code is copyrightable.

Indeed in the contents is this:

> While some prompts may be sufficiently creative to be protected by copyright, that does not mean that material generated from a copyrightable prompt is itself copyrightable.

In usa (but not uk), effort isn't relevent, just how "creative" the work is.

For example, digitizing famous paintings is labour intensive but not considered copyrihtable.

That is the "sweat of the brow" copyright argument, and it has been rejected in by US courts.

Reference: https://en.wikipedia.org/wiki/Sweat_of_the_brow

Rejection by US courts: https://en.wikipedia.org/wiki/Sweat_of_the_brow#United_State...

The parent is complaining of amount of effort put into the work but should be justifying that each of those steps was a creative decision by a human which contributed to the end result.

With ControlNet and LatentCouple you are doing straight-up set direction. With Loras you are casting, costuming and directing the film/lens/lighting.

One could argue that setting up a printing press for high-quality art printing is creative. Deciding what inks to use for a 7-color press and how to separate the original image was a creative act.

Or at least it used to be. Now it's automated.

the OP doesnt include any exceptions based on effort, so no.

> If I’ve spent 8 hours getting the perfect prompt, fine tuning a few LORAs, mixing them, choosing between 8 different checkpointed stable diffusion models, and have done a bunch of in painting, does this constitute a copyrightable work?

The 8 hours doesn’t matter. By the USCO’s ruling, iterative prompt refinement probably wouldn’t, but that’s not super clear (partially, because the USCO’s description of what the model is doing justifying its determination is a fundamentally strained metaphor, so while its clear how it applies to the exact case it describes, its not clear how it generalizes.)

If you are fine-tuning a model (LORA, Checkpoint, whatever), then your input isn’t just a prompt to the model (once or in an iterative process with review of the output), so, your pretty far outside of where the ruling provides clear guidance.

> and have done a bunch of in painting,

inpainting is probably the thing most (even though it again involves prompting, it involves specific selection of where within the image to apply that based on aesthetic concerns) similar to the traditional creative parts of visual art, and the strongest argument given the shape of the USCO description of its rule.

But, I have a feeling that that a rule that appeals to tradition and vague analogy to lower tech visual techniques probably won’t be anywhere close to the final word on copyrightability in this space.

IANAL but I'd assume "inpainting" is legally considered equivalent to arranging. You're not pushing pixels in Photoshop, you're just running the AI again on a particular part of the output, thus combining that output with previous output. This seems equivalent to bricolage at best.

That sounds like your prompt is protected by copyright. I don't see why the output of a machine you feed your prompt into should be.

That argument runs afoul of the "rule of the shorter term" of the jurisdictions that start the copyright term from the publication instead of creation.

Very interesting take! I wonder if a similar argument would apply to something like https://www.youtube.com/watch?v=ZOD7HQOnKAE - where the setup is a lot of work, but the actual execution is entirely up to nature/gravity.

The big clue here, is that if we allow generated content to be copyrighted, then some asshat company will just generate All The Things (within, say, a given space) and copyright them (likewise for trademarks and patents, which could be much easier to exhaust within defined spaces). Combined with the lifetime of companies and the preposterous copyright extensions granted to them, swaths of human-generated work could be excluded from copyright, blocked by some hidden trove of copyrighted machine output. Especially bad if someone manages to copyright everything that could be generated by some AI.

Unpopular opinion : the government shouldn’t protect your art from getting used as training data. The government shouldn’t protect your generated art from being copied. Intellectual property is not real property. Nothing is being taken from you. Force is only justified in response to force, and you don’t get the right to throw someone in jail for “stealing” something you still have.

Do you also agree that government should not enforce if I borrow your bike while you are not using it if I return it? If not how is that different? You still have the bike and you werent using it while I borrowed it - you just didn’t know I was borrowing it

Probably doesnt mind that entire countries have built their economic models around copying ip and selling copycat products.

In universe of perfectly slippery spherical cows, I wouldn't mind. However, in this universe bikes experience wear, I wouldn't be able to spontaneously decide to use it, etc.

You writing a shitty knock off Harry Potter novel "wears" on the brand. Just think your favorite book, movie, or music and think if you would enjoy it if there was really no way to know what was real work and what was just some copy-cat or straight up scammer selling rubbish. Now it is easy; you can not publish your Harry Potter novel due to copyright.

Put it in another way: to become a author in a world where copyright is not a thing is pretty much impossible. Either you have to sell your book digitally with draconian DRM - and you will still get your book stolen or you have to own your own print shop, because if you send your book to any publisher they can just take your work and publish it as their own.

And let's be real the amount your bike wears out from me riding it to the shop and back is so negligible that it won't make a difference in the life span of the bike.

There are good reasons to have copyright - however I do not like how long the copyright is. I've stated it before that in my mind good copyright would be something like 10-20 years or life time of the author whatever comes later. This would allow any creative to hold the right to their world/characters/whatever until they are gone and it wouldn't discourage them from publishing in their old age since even after they have passed their families would still benefit from the works for sometime.

I agree with the business use-case for copyright, trademark, intellectual property, it is a useful collective agreement. To the extent that it is useful.

I thought about this a bit more, and I agree that it is useful that giant software automatons and conglomerates should not get the benefit human creativity until they pass down the value to the creators. Automation should mean less work for everyone, not hyper-profiteering for the few.

My bike is real property. Your thoughts are not.

Tell me you have never done anything creative in your life without telling me you have never done anything creative in your life.

Implying I’d be willing to bend my principles when they don’t benefit me, which I wouldn’t. That’s what makes us different.

Press X to doubt

Should you be fined for copying an NFT or do your intellectual “property” rules only apply when they benefit you?

What does "copying an NFT" mean?

1. You mean copying the image that the NFT points to?

    The copyright of that image still exists even with the NFT. So existence of NFT doesn't change anything about the picture's copyright. With or without NFT you can not sell prints of the image or use it however you want without permission of the copyright holder.

2. You mean copying the actual NFT i.e. somehow taking over the block in the chain and assigning it to you?

    That is a kin to stealing. Why does it matter if the property is digital or physical? This is the kind of mental gymnastics kids use to justify their pirating of entertainment. "I only *copied* this album, I didn't steal it, the artist still has theirs" is such 14 years old's take on the issue. If you don't want to pay for the media you are consuming then don't pay, but then you also shouldn't consume it. This kind of take just shows more how some people haven never created anything of value, yet they want to extract every bit of value from the society.

It sure would be nice if we lived in a world where copyright and patents weren't necessary, but people try to take advantage of everything and anything, so to protect innovation and creativity we need copyright and patents.

There is no such thing as private property. It's a fiction enforced with the threat of violence. It's even fairly recent as a concept.

If we agree that private property should exist, "it's not real" is not a good argument for why intellectual property shouldn't exist. Not all violations of private property have to be stealing. If I sleep in the empty house you own, that is trespass even if I don't cause any physical damage. If I tend to the garden of the summer home you own and eat the fruit that would have spoiled by the time you came by, I still violated your property rights even if I left the garden in a better condition than I found it.

If we abolish intellectual property, why should we keep money, which at this point is entirely virtual? Why should we allow stocks, which represent a partial ownership claim in a legal entity that lays claim on other property? Why should we allow corporations, which are afforded similar rights as real persons but don't physically exist? Why should we allow ownership of land which is unused, homes that aren't lived in, produce that isn't consumed? Why should we allow private property at all?

If you want to abolish private property, I'm all on board with you, but if you think intellectual property is tangibly different you need to take a closer look at how private property came about and what it even means.

Private property !== personal possessions. I, for one, don't agree private property should exist, just like intellectual property. There is no place in the future for such concepts, they all feel quite antiquated.

What about possessionless property rights, that are only utilised to enforce theft and robbery laws?

Am I physically touching your painting when I copy it? If not, then your attempt at equating physical property to thoughts is nonsense. We both believe in physical property, however arbitrarily invented it is. Only you believe in intellectual property, and I’m willing to bet you aren’t even consistent with it. Should you be fined for saving an NFT?

I don’t want the government to be involved in virtual money. Stocks are a claim to physical property, corporations are (or should be) a convenient way to address physical property pooled together from multiple people. The rest of your comment can be addressed with the homestead principle - if I didn’t take it by force you can’t take it from me by force.

So if I trash your house that's cool? You still have your house, it's just in a different entropy state.

Or maybe one particular state had more value than another?

So maybe when other people value different entropic states over others, perhaps that has value?

> Am I physically touching your painting when I copy it? If not, then your attempt at equating physical property to thoughts is nonsense. We both believe in physical property, however arbitrarily invented it is. Only you believe in intellectual property, and I’m willing to bet you aren’t even consistent with it. Should you be fined for saving an NFT?

I'm interpreting your view as property only applies to physical things or abstractions of physical things. Is there a name for this theory? I'd say I subscribe to the labor theory or property and it's interesting to think about this from another perspective.

There's a couple cases I'm having trouble with. For example, is an ip address property? Seems so because it represents a physical scheme about the routing of ip packets. Similarly an email address or physical address seems like property. If any of these were replicated/spoofed by someone else, they would disrupt the ability to physically route whatever (bits, packages). Ok.

But then what about trademarks? In some ways those are 'pull' addresses - I go to the store, and buy a Coke and I'm expecting that thing labeled Coke to be the physical product I expect, not some spoofed version of Coke. But that label and what not is just a picture, and the original picture wasn't harmed in any way when the spoofed version was created. Is this concept just not allowed and we accept that inefficiency? Or maybe trademarks are allowed as property because they represent the physical concept of the systems the Coke corporation has put in place to create and produce Coke.

But if that's the case, then what about a non-trademark painting? If I go to the store and see a painting by an artist I like, and want to support them and the physical systems the went into generating that painting, and not be spoofed should that be supported?

(Though maybe all this could be solved with a QR code or DOP label or something. Though somewhat inconvenient to have to robustly check that all the time).

Honestly, it seems like if you accept any contract law you implicitly must accept copyright law. Consider the following contracts:

I agree to harvest your crops in exchange for 50% of the yield.

I agree to sing if you give me supper.

I agree to sing if you give me supper and not record me.

I agree to sing if you give me supper and subscribe to all the rules of copyright law.

I agree to sing if you give me supper and subscribe to all the rules of copyright law and all people in your jurisdiction also subscribe to copyright law. (And then some folks who wanted a singer got together and passed a law that satisfied the singer's requirements).

All these seem reasonable to extensions of the previous and it seems like we've ended up at the latter.

You might say 'well I didn't agree to that contract', but there are plenty of things in the world that we are bound by that individuals that we didn't explicitly agree to. I didn't explicitly agree to not dump motor oil in the ocean, or hunt deer on a certain day or that 50 decibels as the bound for noise during the night, or that children of age 12 can't vote, or that I have to wear pants in public.

You might argue that the last step is invalid, and instead the world should be a place where individuals can pick and choose which laws they are interested in subscribing too. I don't really want to put much effort into that path because my intuition is that's equivalent to anarchy, but maybe there's an alternate argument here.

So it seems to have a world w/o copyright law, you have to throw out contract law, or convince everyone to stop somewhere on that chain, perhaps not to make the contracts that mimic copyright law, or make a law making such contracts invalid. The later two options seem fairly artificial, impinging on folks self determination. So we're left with having to eliminate contract law, which seems pretty limiting.

I want to hear you respond to this, as I'm about to respond to you. I see no reason to continue if you're just going to avoid anything I say while yourself trying everything to see what will stick.

It's not alien that I believe property excludes thoughts, as you're trying to make it seem. Most people agree with me, until it wouldn't benefit them to do so. Again, I demonstrate this with your presumed support of IP for artists and opposition to IP for NFTs. You would argue against IP for NFTs the same way I'm arguing against you right now, the only difference is that I'm consistent.

IP addresses are not property. Neither are trademarks. We can argue about the slightly related topic of whether they deserve government protections anyways, but it wouldn't be because they're property.

No, copyright and contract law have little to do with each other and you're making a giant leap of logic. Namely, I can find a way to obtain your copyrighted content without making any agreement with you, and still be held legally liable for it. Torrenting copyrighted content someone else broke the rules to seed is one such example. On the other hand, I can't be punished for listening to you sing for someone else in exchange for food, even if I didn't give you any food myself.

Yes there's plenty of things that we don't explicitly agree to and I'm in favour of keeping that list as short as possible. I don't agree that some of the things you listed should be implicit. Nobody is perfectly consistent, but at a baseline I'm far more consistent than you are.

Re nfts, I'm kinda not clear on the question, I think because we don't have a clear societal statement of what an nft is or is trying to accomplish. Literally what an nft is an entry containing with a url on a Blockchain with certain properties around non fungability. Saving that has almost no issue because distributing the bits in a blockchain is one of its fundamental purposes. Now saving the file referenced by the nft starts to become more interesting. My interpretation is that we should respect do whatever the rights holder wishes. If they want to allow saving, great, if not, then don't save. The problem seems to be the rights assignment of nfts aren't really communicated or standardized afaik. Some feel like the nft should represent the ownership rights of the work, some feel like they should represent the ownership rights of an instance which may or may not transfer the rights of the original owner. Some feel it should have nothing to do with rights transfer and instead just be it's own thing. Any of these stances (or others) are fine with me, and if any disallowed saving that seems totally fine. It would be nice if whatever scheme/contract was either standardized or easily identified for each work/nft. Images on the internet already have this problem and systems like creative commons attempt to help, something like nft-commons could really help make clear what rights are implied by the nft and thus which images referenced by nfts could be saved.

So re ipaddresses what characteristic differentiates the from property? For them to be useful it would seem they must have pretty much all the same characteristics.

I agree there could be a world where people could torrent whatever w/o repercussion, in the same way there could be a world where we didn't have to wear pants. It's not where we are, but we can imagine and reason about it.

But even in such a world, should a singer not be allowed to make a contract where they sing but won't be recorded, or a contract where they are recorded only if the folks follow copyright law? Effectively isn't this what many trade agreements are, one country will give another favorable trade conditions X in exchange for Y which includes 'your population can't torrent'. The country voluntarily does so because X is worth it. In fact elsewhere in this thread someone said exactly this happened in their country. Should this contract not be allowed, why not?

So I feel like even if you started in a world that allowed torrenting, if you allowed contract law, you would inevitably evolve to a world that disallowed torrenting, because that's a contract some people would want.

How does this affect programming? Does this mean if a program was built using GitHub Copilot, that in order to have copyright on software, you have to explain that to the copyright office?

Maybe similar to how software is affected today by open source, by providing attribution.

The human authorship requirement still stands:

> If a work's traditional elements of authorship were produced by a machine, the work lacks human authorship and the Office will not register it. [0]

Even with that, applicants now must disclose the inclusion of AI generated content and highlight which parts are human authored vs AI generated:

> Consistent with the Office's policies described above, applicants have a duty to disclose the inclusion of AI-generated content in a work submitted for registration and to provide a brief explanation of the human author's contributions to the work. [1]

[0] https://www.federalregister.gov/d/2023-05321/p-44

[1] https://www.federalregister.gov/d/2023-05321/p-59

If an author chooses not to credit an AI, how are they going to know?

It's already not completely obvious with the current state of the art in at least some domains.

What happens when the tech moves from "Not completely obvious" to "Impossible to tell?"

Right… I hope this isn’t a situation where everyone has to be dishonest in order to compete at the highest level. Major League Baseball and cycling from the early 00s come to mind. I wonder if having correct rules that are unenforceable do more harm than good.

Fraud has existed for a long time.

For the most part society is designed with the assumption that most people will tell the truth.

You could create a different kind of society, where the default assumption is everyone lies, but I suspect no one would be able to live under those conditions.

Note this is not the same as taking steps to check for lying, it's just a question of what's the default assumption.

“Not going to know” can apply to most laws.

Most people especially enterprises abide.

> If an author chooses not to credit an AI, how are they going to know?

USCO doesn't generally proactively investigate (it would take too long), but a copyright claim could be invalidated if it is proven that the applicant didn't disclose required information. As of today, the USCO has explicitly say that AI-assistance needs to be disclosed.

The correct response would be the end of digital intellectual property. We would have a creative explosion akin to the age of free sampling in music, the early internet or the modern Chinese digital landscape. The actual response will be more layers of bureaucracy, which will have the chilling effect of demoralizing small creators and further empowering IP trolls and large corporations with lots of money for this sort of thing.

Watermarking seems like a possible solution: https://www.nytimes.com/interactive/2023/02/17/business/ai-t...

I think that OpenAI et al are incentivized to pursue watermarking. If someone uses GPT to write a best-selling novel or a blockbuster movie script, OpenAI would want a piece of the action.

Similarly, publishers/distributors of creative works are incentivized to use any available detection tools because they don't want to be surprised when someone comes along and says, "Actually, you owe us a boatload of cash for that work."

It's stupid to create a rule that's both unenforceable and limits protecting unique creations generated by AI but initiated by humans. Results are all that matter, not process.

> If an author chooses not to credit an AI, how are they going to know?

IF a dispute arises, it will be settled in a court of law, with the trier of fact (jury or judge, as may be) applying the civil preponderance of the evidence standard. (Tools for detecting use of generative AI models are being developed, as are systems of including watermarks that are unnoticeable by casual human inspection in the output of such systems.)

I've seen Grammarly commercials.

What if I write some paragraphs, and then drag a big Grammarly slider across it, and it's no longer my words, but my ideas are still there, just buffed and touched up professionally?

What's "AI generation" anyway?

I wrote a limerick for a friend last week. Well, I had the idea for it and it was jangling inside my head, but I didn't feel like fleshing it out, so I had the AI write it. I was accused of "cheating". But I'm capable of writing this limerick; I just wanted to see if a computer could put a ribbon on it. And it worked fine. I claim authorship (and copyright) nonetheless.

I would like the Copyright Office to delineate just how much effort is necessary for a work to possess “the traditional elements of authorship.” I type in MIDI instructions and play back a tune, but if I type in sentences it doesn’t count?

Maybe they can measure entropy. Did you emit a sufficient amount of data to make the Music Box work, given the Copyright Office defined ratio?

For every unit of data in your song, you must have emitted no less than 1/2 of a unit, which was specifically intended to manipulate the Music Box in order to shape outputs. Any less and the creation isn’t yours.

What do you mean I can’t copyright this song? My air drumming against the cassette player introduced very significant data, which adjusted the playback. Just a very slight legato. It’s too similar? I assure you it is not. Convert the two to Universal Data Format and check the Hamming distance. Incredibly different pieces of data, believe you, me.

This is basically the origin story for how the US Copyright Office became the galaxy’s Entropy Police. “Back in 2023 someone tried to copyright a photo of an “ottercat” surfing a DeLorean on the moon and now we’re in charge of how organized is too organized.”

as you've discovered, really it just depends on your lawyer

It's always been all about the lawyers, juries, and judges.

The laws have always been the least significant thing in the legal system.

Here you go: https://www.copyright.gov/comp3/chap300/ch300-copyrightable-...

308.2 Creativity

"A work of authorship must possess 'some minimal degree of creativity' to sustain a copyright claim. [...] '[T]he requisite level of creativity is extremely low.' Even a 'slight amount' of creative expression will suffice. 'The vast majority of works make the grade quite easily, as they possess some creative spark, 'no matter how crude, humble or obvious it might be.'"

310.7 The Time, Effort, or Expense Required to Create the Work

"When examining a work for original authorship, the U.S. Copyright Office [...] will not consider the amount of time, effort, or expense required to create the work. These issues have no bearing on whether a work possesses the minimum creative spark required by the Copyright Act and the Constitution."

Note that this is not true in every country's copyright system (some do recognise effort), but it is true in the US.

If I wave my camera around taking lots of pictures, and then find one that I really really like I can copyright it. If I generate a bunch of AI images and find one that I really really like, I think I should be able to copyright that too. In both, the creative spark comes from discerning what has value.

Careful with your wording: you don't generate AI images, the AI generates images. You use the AI to generate images. An argument could be made that the AI acts like the camera and the selection process is the creative act. The text hints at this when it mentions curation. But curation also provides a far more limited protection than authorship (e.g. consider the copyright of phonebooks).

Generally agree. But if it’s the AI generating the images then it’s also the camera generating the images.

There is a precedent that any work done in a data center on behalf of a user, because a user initiated the action, has the same protections as if the user had performed the action themselves. It has been used, in connection to copyright, to allow "cloud DVR services".

I'd argue that in the case of AI generated work as we see it now, especially if a prompt from the user is required, that work would not exist except for the action of the user, and is thus created by their action.

Would this not apply to tons of procedural generation tech as well (used in plenty of games and tons of movies)? We give it a bunch of elements and write algorithms to do the generation... but we are ultimately not the ones putting pen to paper so to speak.

I think this is the 100% the right call, some small level of human effort should be required otherwise what is to stop a few individuals from mass copywriting 10 million images and suing everyone that produces something substantially similar.

Having said that, there may be odd knock on effects at play here.

Not possible to reliably tell apart. The Image generation scene is one thing. For a lot of output, there are still some tells. Text generation though...well good luck ever finding that out.

It sounds like the honor system, which is not unprecedented. A lot of governmental systems rely on people to tell the truth when interacting with the government, and then they have to police cases of suspected fraud after the fact.

If you were dragged into a copyright suit by someone, you could probably try to prove that the work was AI generated as a defense? Not sure how effective it would be.

> If you were dragged into a copyright suit by someone, you could probably try to prove that the work was AI generated as a defense? Not sure how effective it would be.

It would probably be similar to when wordperfect was sued for including famous painting as clipart. WordPerfect won using basically this defense (that no human creativity went into making the clip art) https://en.m.wikipedia.org/wiki/Bridgeman_Art_Library_v._Cor....

Government just found a way to expand itself and give out wages for a job that they cannot really do.

Who has the burden of proof? Can I just claim everything and anything as “made by AI” and start (ab)using it and then it is up to the original author to proof that it wasn’t actually made by an AI?

You do, in a civil claim its up to you to prove that your rights were actually violated.

It’s going to be insanely hard to copyright anything produced by AI. Especially now that generative models are widely available/in use. The possibility of multiple instances of the same output being produced (as a whole, or in parts) is high.

All you have to do is set the temperature low and the generative models will start producing repeatable results.

All of a sudden, copyrighting GitHub Copilot supported source code is not so straightforward anymore. How does one distinguish the human authorship in a function from that generated by AI?

Interesting times we live in. In a way, this may lead to less regulation/more openness. Equal opportunity for all. At least one can hope.

I like to watch people play with Midjourney in the public Discord channels, and I can only assume that if the AI generated an image for somebody, it could generate the same image for me.

(If the seed and prompt just happen to be exactly the same)

People using AI generated images are not going to be able to sue for copyright infringement because they can never be sure the AI didn't just generate the same image for somebody else. (and more importantly, whose image was generated first :))

Exactly. I thought about this some more. There is a way.

If the training data is unique, and the copyrighter can prove that, then I would consider the end-result copyrightable.

Also - what’s stopping people from lying on their applications about the origin of their work?

It's all well and good that the copyright office deems that protection only applies to the parts of the work where a human was in control of the traditional elements of authorship, and that there is a duty to disclose the use of generative technologies ... but now one can plausibly assert that they are the sole author of a work which was produced by one of these models, and be 100% confident that no one can dispute that.

For a single work, yes. If they turn it into a high-volume business, it might become more difficult to maintain the plausibility in the long run, and they could be asked to put their claimed artistry to test in a controlled environment.

Being the copyright owner and being the author of something are very different things.

And it doesn't matter because anyone can use the AI to produce thousands of similar images with little effort.

All those tools have randomization so they don't produce the exact same image for the exact same prompt.

The value of an image produced by AI is very low because the cost to produce it is very low.

No point in committing federal fraud by falsely claim you made them.

> no one can dispute that

Of course I can dispute that.

I can even prove you didn't make the AI image by asking you to produce a work of similar quality under supervision.

If all you can do is type AI prompts into a computer, good luck producing an actual painting.

The courts are not going to be able to stop the flood of AI generated images.

Even now, when it takes a lot more human effort to generate an image because It has to be done by a human, the courts are ineffective at stopping the rampant and blatant copyright infringement that happens daily on a truly massive scale on the internet.

Now that the floodgates of creativity have been opened by AI, the amount of content being generated is going to exponentially increase, and all the laws and courts in the world are not going to be able to even slow it.

> I can even prove you didn't make the AI image by asking you to produce a work of similar quality under supervision.

Have you ever seen a court do anything even remotely similar to this other than in Hollywood movies? Not only would there be serious freedom of expression/compelled speech aspects to doing that, it would be entirely impractical and ultimately prove nothing.

This announcement has immediate, significant practical impact for creatives.

The most important document a creative had, up until today, was their portfolio -- typically, a look-book of finished pieces.

Now, that portfolio needs to include, for every piece, proof-of-work -- snapshots of the whatever-it-is in various states along the road to completion, in sufficient quantity to dissuade any legal claim that the work was AI-produced.

At the limit, those aim to sell organic-certified free range content will want to surveil themselves during the entire creative process, and associate that recording indelibly with the created work.

That's a whole lot of extra work and a whole lot of extra privacy violation, but the alternative will be to devalue one's own work: for to the extent to which it may have been produced by AI, it will be a liability to downstream consumers.

For example, a film director might commission a score from a composer, but unless that score comes with timestamped, SHAsummed video of enough of the composition process to preclude invalidating the broader claim to a copyright (and thus salable work), that composition becomes a financial risk for the director.

The consequences of not doing so are severe: At minimum, if the score cannot be copyrighted, then it can be borrowed, free-of-charge, by another film director, and at maximum, the spectre of AI contribution might virally taint the entire film (IANAL; am I getting this right?)

"Creative" just became the most surveilled job on the planet.

We also may have just found the first agreeable use for blockchain -- an indelible public record of organic artistic creation, bearing SHAsums associated recorded twitch and youtube streams (along with logs of workstation network traffic) permanently with the finished work.

Wouldn't it be hilarious if AI copyright law saved crypto?

"at maximum, the spectre of AI contribution might virally taint the entire film" - which would also mean that anyone who used GitHub Copilot would not be able to copyright their software code.

I hadn't thought of that -- but yes, it very well might be the case.

Let's say there's a 50% (extremely generous!) limit on AI contribution.

An efficient company will operate close to that line, say, targeting a 49% AI contrib. (Remember how inexpensive AI contribution is in comparsion to human contribution. Market pressure will push people towards tenths of percentage points.)

That means that Joey, your new intern, might accidentally commit enough SLOC to make the whole codebase uncopyrightable.

Even if Joey didn't do that, the possibility of Joey doing that will have a strong deterrent effect on hiring Joey, especially if he's WFH.

The only mitigation would be to record everything Joey does, so that if there's ever any question, they can whip out the recordings of Joey picking his nose or whatever.

Something I've begun thinking about after playing with GPT-4 is that I could see within a year or so any of my new software development projects being a "no code" setup using comments to define functions and correct problem areas but leaving all actual code to the generative AI.

In many ways software projects for years now have this issue.

A decade ago when ruby on rails scaffolded out a project - is the result copyrightable according to the new guidance?

Separating out copyrighting software design at a comment level from software implementation is going to be the direction this all goes as the tools rapidly get significantly better.

Which is also going to be great, as imagine how a codebase designed this way might be able to be switched to a new language or switch out the 3rd party API being used or database being run on.

People are worried about protecting their busy work rather than evolving with the technology to establish their value above and beyond the busywork parts.

My value in software engineering isn't in typing up the loop, and less even in knowing that I'll need one. It's in knowing how to manage complexity across a broader cross integration of concerns. The way I'm doing that will change as technology advances, but it will still be some time before that part is automated too.

> My value in software engineering isn't in typing up the loop, and less even in knowing that I'll need one. It's in knowing how to manage complexity across a broader cross integration of concerns. The way I'm doing that will change as technology advances, but it will still be some time before that part is automated too.

I think this is just normalcy bias on your part. The thing that GPT-4 does apparently works at most levels of complexity, certainly more than GPT-3.

It seems likely that, for sufficiently high N, GPT-N will enable your boss to, as you say, "manage complexity across a broad[er] cross integration of concerns".

There's nothing magical about "cross integration of concerns," any more than there was something magical about being able to use (say) `git rebase` correctly.

There is no reason to think that GPT-N won't be better at you for that, too.

Why would your boss pay you six figures when she can get the same deal for $20 a month?

So if movies use AI to help create characters they won't be able to claim copyright? I expect Hollywood studios will have something to say about this. With AI already being deeply integrated into DCC tools this stance isn't going to last long. And how will it be enforced if the artists don't show how they made the art?

There's still a ton of human expression in the average movie, so they're pretty safe. If they edged towards being fully AI though I could see your concern coming true.

Applying color balancing in photoshop doesn't make the original image taken by a photographer the product of a machine.

Laws are not black and white.

The whole point of lawsuits and lengthy opinions by judges is to have a nuanced interpretation of the law.

If the whole movie would be generated by a computer from a single sentence then yes, it wouldn't be copyrightable.

But as long as producing movie involves human screen writers, human directors, human performers, human camera operators etc. they are obviously creative works of those humans and therefore copyrightable.

It's great that USCO is ending speculation with an official stance. Their rationale seems perfectly reasonable, and in step with the court rulings that they cited.

That's not going to be practical. How do you distinguish if something was created by mere computation, or by AI? Oops, AI is just mere computation. So if you forbid mere computation, I cannot use fancy numerical algorithms in shading my 3D art? Ain't gonna fly.

One part I found particularly interesting about this is footnote 27:

> While some prompts may be sufficiently creative to be protected by copyright, that does not mean that material generated from a copyrightable prompt is itself copyrightable.

I wonder if this is meant to imply that no possible prompt can produce copyrightable output. What if the prompt already contains expressive elements that the model faithfully reproduces in its output? (For instance, in the limiting case, "Exactly repeat this poem that I wrote: ...") Or perhaps this is meant only in the strict sense, that a creative prompt does not necessarily produce copyrightable output, if the expressive elements in the output originate from the model instead of the prompt.

There will be a case-by-case weighing of which artistic aspects were contributed by the human and which by the AI. What they are indeed saying here is that a prompt being copyrightable does not by itself imply that the generated output is copyrightable. I believe this also means that if someone takes your copyrighted prompt and uses it to produce some output, then that output doesn’t a priori infringe on your copyright.

So this effectively destroys any concept of AI digital art for sale. If there’s no way to protect it then there’s really nothing to own. It would be like trying to sell someone public domain content.

There's plenty of copies of public domain works available for purchase on Amazon. It does mean that AI digital art will fetch commodity prices, but not that it won't be sold.

Sort of like hotel art.

Not quite; having it exist is something that may be worth paying for, even if it's equivalent to public domain.

As an AI artist who's been selling digital art, it really depends on whether you really care about "protecting" your work. I've seen my stuff stolen all over the place, NFTs minted, etc. Most sites will take the copycats down. Lack of copyright protection doesn't necessarily supersede individual marketplace policies.

"I've seen my stuff stolen all over the place, NFTs minted, etc"

The overwhelming majority of artist do not have the resources, time, or inclination to constantly chase down copycats and copyright infringers.

Copyright law is just ineffective at stopping or even penalizing most of the infringement that goes on in the digital era.

It's a relic of a past when making copies was hard.

Digital art as a whole has already been made worthless by the creation of diffusion models, whether there is copyright on them really doesn't matter at that point, because pretty much any image that can be imagined is now trivial to produce in under a minute.

"pretty much any image that can be imagined is now trivial to produce in under a minute"

That's overstating it a bit.. but it is likely we'll get there within the next decade or so.

Currently there's still a lot of art that's hard for AI to mimic effectively, and only certain types of art can be created quickly and without much effort using AI.

The above is based on my experience of creating well over 5000 images in Midjourney, and keeping a close eye on AI generated art created by others.

It's amazing tech that's close to magic, but it still can't do everything.

You're neglecting the fact that the types of art that can be created quickly are the ones that make up the bulk of the training data, which are also the types of art most people will want to make.

> pretty much any image that can be imagined is now trivial to produce

What? That's not true at all!

If you're talking about "art", most artists are fussy and want things exactly right. They will spend a long time tweaking the image, or combining different images, painting, processing, pixel-pushing to get the thing they imagined.

If you're talking about some random need for an image of a "horse with metal legs", and settle for whatever the AI spits out, and then call that "my valuable art", that's different.

There is plenty of ability to tweak AI images, I don't understand this objection.

You claim "all digital art is worthless because it can all be tweaked into existence in under a minute".

If that were true, I could show you any piece of digital art, be it 2D or 3D rendered, and you should have no trouble making an identical copy in under a minute using only diffusion models and tweaking! That is an impossible task, I promise.

I'm enjoying stable diffusion as a fun playground and rendering engine. But it can't magically produce the exact picture in my mind.

Even simple things like "hero tree in dark misty forest, old swing hanging from large twisted branch, path extending into gloom". The AI will produce a nice image, but not the one in my mind. The image in my mind is important for style or composition reasons relative to the whole work. Artists rarely think in terms of isolated one-hit-wonders, they are producing work with important and meaningful connected threads and detailed elements.

On a practical level, in order to protect authors and inventors (i.e. give them a reason to keep doing their work), this makes it even more vital to closely lock down AI interactions and preferably bring models offline in the sense of running locally on hardware without network component and where the bits are not sent anywhere except to the user. That way the output of the interaction can still be covered by trade secrecy if not published, since it will not be covered by copyright law if published or leaked by employees at the hosting company who have access to the output of the interaction.

To give a specific example, if as a paying user you craft a prompt to give you a competitive advantage, it will not be protected by copyright if leaked.

You can still run the code on your own server and others can still try to guess how you did it, but they should not have access to the interaction unless you explicitly publish it.

This will continue to encourage innovation. It is not necessarily the best trade-off but it is understandable.

Just curious... with such a decision in place, wouldn't more artists be inclined to make their use of generative AI less transparent? In other words, what would prevent them from ceasing to disclose that their work includes elements generated using generative AI?

This could indeed increase skepticism, particularly when it comes to img2img and, if not as much, txt2img usage.

Exactly. When someone rips off someone else's art a court can usually determine that the copyright was infringed. But when an artist uses an AI model trained on other people's works how can that be proven even in a courtroom?

Hmm. I am not so sure about this. What about the fact that to produce a image I like with dalle our similar I need to often run many many different descriptions to get what I want and some times use the edit feature. If a generate an image and use the built in photoshop like erase feature and regerate something else to fill the space is it copywriteable. What about if photoshop has or implements (probably already exits) auto insertion of images and people. Say you just paste something onto your image and photoshop runs ai algorithm that helps blen it in. What about traditional painters that relay on randomness in thier work? Like guys who just spin a canvas get blindfold and just throw paint around?

Then with DALL-E it is like you are instructing a commissioned artist (who would own the copyright). I think it's fundamentally different to using a tool in Photoshop - content aware fill uses the rest of the image to fill in space, so whoever owned the original image would own the content-aware fill section. If an AI is used to generate new features - images, people - then it is like a collage or original and copyrighted work. Artists that throw paint around made the artistic and creative decision to do that, would choose the paint, the canvas etc. and actually it is very difficult to create work that has that quality of spontaneity and randomness - they would definitely own the copyright.

I would speculate there will be no works containing no material generated by artificial intelligence soon. At least it will become a common practice to use a GPT to improve your language, possibly also assist the creative proccess by supplying facts and suggesting creative ideas. Everybody uses spell checkers and thesauri already (some also use grammar checkers like Grammarly), also google things up - this will be increasingly extended/replaced with advanced AIs usage. It is generally considered correct for a writer to hire assisnatnst who would proofread their writings before publishing them or even assist their thought through conversation so I can see no big deal in an AI doing the same job.

What the government needs to do is to say that all AI-generated content is CC-SA 4.0 or GFDL and all the code is AGPL <latest version>. That oughta do it.

(edit: I put "CC-BY-SA" and then I remembered there's no need to give credit to a hunk of sand and electrons!)

Or, specified more correctly, a share alike and attribution regime not based on copyright.

Reading this, it’s clear a tool like InvokeAI (built on Stable Diffusion but with potentially a lot of human input in a mouse-driven GUI) could certainly qualify for human authorship under their principles.

In the studio, the photographer directs the lighting, the model, tweaks the shot, issues commands to modify the scene, presses a button, checks the outputs, iterates, chooses the best shot.

In a virtual conversation, the promptographer directs the computer to set the tone, chooses their model, tweaks the inputs, issues commands to change the virtual parameters, presses a few buttons, checks the outputs, iterates, and takes the best output.

One of these is a creative work guided by a tool and one is a act that shows no human creative authorship at all.

Creating AI artwork is simply telling the computer the result you want, and picking what you believe to be the best output. There may be something of an iterative process there, but it's really not creative.

Also the photographer is not "guided by a tool". The camera doesn't guide them.

Can you elaborate on what makes something creative or not?

On the 2nd point, the camera guides photographer. Watch anyone take a selfie, they use the camera to decide what to do with the camera. It is iterative.

Sure. It's like having a friend who can draw. You detail to them something you're picturing in your mind, and they make it into something actually real.

Who would have ownership over that art piece? Who actually did the work to make it exist?

Copyright doesn't protect ideas, it protects the specific creative expression of those ideas. By creating a prompt you're creating an idea, but that isn't a copyrightable, creative action. You're just the 'ideas guy' for a robot artist.

And again on the 2nd point, the camera doesn't guide the photographer. It can help them dial in their vision, but they're in full creative control at all times. It guides you no more than a hammer guides you to the nail.

Good. Works containing any material should not be copyrightable.

>For example, if a user instructs a text-generating technology to “write a poem about copyright law in the style of William Shakespeare,” she can expect the system to generate text that is recognizable as a poem, mentions copyright, and resembles Shakespeare's style. [...] But the technology will decide the rhyming pattern, the words in each line, and the structure of the text.

Neat example and convincing reasoning.

If this is the logic, then corporations should lose personhood because they can't have it both ways.

I don't see the rationale because ultimately there are owners of copyright, whereas the creation process should be immaterial to protection whether it was paint thrown against a wall, AI generation, or some poor human painting with hummingbird eyelashes.

> whereas the creation process should be immaterial to protection

On the contrary, the US position has been for a very, very, long time that the creation process is the only thing that's material to protection. The result is not the thing that matters.

Corporations can own copyright not because they're human beings, but because the copyrighted work was originally generated by a human being's creative expression. There's no contradiction or having it both ways here: this isn't a set of guidance about who can own copyright, it's a set of guidance about what can generate copyright.

This works for now. However, I can imagine a spectrum of input parameter specificity that could eventually blur the line this draws and undermine this strategy.

On one side you have these general prompts.

On the other side you have, say, autotune, where the final output is still created by a machine but the human's input (voice in this case) greatly constrains the output.

Pointless screaming at the tide hoping it won’t come in.

In 3 years there won’t be a major digital tool that isn’t pipelining through ai especially in CG and music, in 10 years the pre-AI versions of those tools won’t run on a modern computer.

Unless you’re ready to say “anything made with Adobe CC can’t be copyrighted “ then this is pointless to debate.

They specifically address works that use AI but are sufficiently modified by a human

Seems like a win for artists who oppose generative AI. Now those who use AI to claim a copyright are "put on notice" to clearly disclaim the AI generated portions of their content.

It will be interesting to see if Prompts themselves could be copyrighted, since presumably humans came up with the prompts.

Recipes are usually not copyrightable as they are just a list of ingredients and instructions with no room for expression. I wonder if the same argument could be applied to prompts.

They state this in the document as well:

> While some prompts may be sufficiently creative to be protected by copyright, that does not mean that material generated from a copyrightable prompt is itself copyrightable.

It sounds like it would depend on the complexity of your prompt.

Companies will just outsource to small studios/individual artists that don't admit they use AI.

If a TV show generates a script using AI in which animated characters read using AI voices... would I be allowed to then distribute/remix the portion of that episode? What does the distinction requirement actually do besides serve as a notice, is there some kind of diminished rights?

In this case they're saying they wouldn't register it, you could still try to take someone to court for copyright infringement. But you'd be rolling the dice since "the copyright office says it's unfit for registration" is probably a pretty convincing defense to the average judge.

It wouldn't be the first time there was an argument over whether a work was copyrightable or who had the rights, though - for example the Monkey Selfie case https://en.wikipedia.org/wiki/Monkey_selfie_copyright_disput...

How come work “autonomously created by a computer algorithm running on a machine” cannot be copyrighted <<it was made “without any creative contribution from a human actor”>> but machine code as generated by a compiler can?

In addition to the new registration guidelines, the USCO announced a new AI initiative and series of public hearings

https://copyright.gov/ai/

Art is in the eye of the beholder.

We're going to be awash in so much AI generated content that the act of lifting something truly marvelous out of that background will be a creative action worthy of protection.

What I’m curious about is how much Neuralink will it take to make a person legally less-than-human in this context. If I’m hooked up to the Metabrain am I still entitled to ownership of my works?

But if you asked it to generate a story in the style of Stephen King, whose works are still copyrighted, then does King have a copyright claim even though he didn't operate the AI?

Now the big question is how do we make “ai” companies pay for the data they use to generate their content. Stealing ip by scraping content and then reselling it is not fair use.

Ok, so I won't tell anyone what tools I used to make my art. Not a big deal.

The problem will come when somebody takes your art and starts selling it themselves. It will be down to you to prove you have a copyright.

ChatGPT sues US Government

Anti AI discrimination NGOs call for public protests.

A non-person cannot sue.

Autotuning of voices uses AI- Songs using autotune are partially AI generated.

chatgpt4 summarized:

This is a statement of policy by the U.S. Copyright Office to explain how it examines and registers works that contain material generated by artificial intelligence (AI). The statement is effective from March 16, 2023. The Office has experience and expertise in distinguishing between copyrightable and noncopyrightable works. The Office may require additional information from applicants who use AI to create or use copyrighted works.

Thats a pretty bad summary

Pretty correct, but it wasn't mine (my comment) but chatGPTs. I just presented the commentt from chatGPT for fun - which might not comply with rules that i dismissed. Sorry, i wanted to publicy show the summarization-feature, but still: it was not my best idea. You can always do your own summ...

Wasn’t there a voyager episode about this?

Please let chatgpt4 summarize this

non ai content will be drowned out by the massive wave of generated content.

And how will they know?

The document is too long let me summarise it with ai

> But the technology will decide the rhyming pattern, the words in each line, and the structure of the text.

But I can certainly specify those things in the prompt. In fact I can write some of the poem and have it riff on the rest for me. And anyway how would you know whether or not and how much I was assisted by the AI. Strawberry Fields forever.

-- Written by llama 13b, edited by a human(?)

  -- Written by a human, edited by llama 13b(?)
    
    -- Written by llama 13b, edited by a human(?)

> And anyway how would you know whether or not and how much I was assisted by the AI.

Because in a lawsuit you would have to reveal that info.

You can't just sue someone and not answer questions pertinent to your claim.

  - Your honor my wholly original work of Halvin and Cobbs, a cartoon of a little boy and his pet tiger is entirely original to me.

  - Did you use an AI to produce it?

  - See, the problem with fine art is that it's supposed to express original truths.

    But who likes originality and truth?! Nobody! Lifes hard enough without it! Only an idiot would pay for it! 
    Popular art knows the customer is always right! People want more of what they already know they like, so popular art gives it to 'em!

  - Sir, did you use an AI to plagiarize your way to an unofficial sequel of a beloved comic strip? Answer the question!

  - Yesn't.

You are then held in contempt for not giving a straight answer.

Its not like no one has ever thought of the idea of telling half-truths or digressions in court as a loop-hole before. Pretty sure judges take a very dim view of that.

> - Your honor my wholly original work of Halvin and Cobbs, a cartoon of a little boy and his pet tiger is entirely original to me.

And as you submit this to the court, the other party requests discovery of evidence from your computer, your communications and (based on that) the API logs of a particular third party provider you could have used, and uses that to not only dismiss your court claim but forward it to the local prosecutor - previously it was just a civil case, but perjury is quite punishable, in USA up to five years in jail (https://www.law.cornell.edu/uscode/text/18/1621).

Soon inference for big LLMs will be ran locally, without leaving any kind of logs which can't trivially be erased.

I'm sure there will be a number of ramifications but one I know of is for those who sell specific captures of guitar amps that were created through machine learning just lost their copyright on that product. Those works could not have existed without the AI that was used to create them.

The U.S. Copyright Office seems to be living under a rock and completely unaware of how much creative work goes into prompt engineering, as well as selecting the subsequent image and applying variations to it until it meets the human's creative desires.

If there is substantive work in the development of the prompt, then the applicant can submit the prompt for copyright.

The finding seems pretty reasonable, if the machine is doing the work then the machine can claim authorship. Otherwise, we would say that the Pope was the author of the Sistine Chapel as they prompted Michelangelo to paint "a large curve display of heaven including with clouds and pictures of angels triumphant".

They state this about the prompt engineering:

> While some prompts may be sufficiently creative to be protected by copyright, that does not mean that material generated from a copyrightable prompt is itself copyrightable.

The prompt is fully copyrightable but the output isn't.

This is already a hot legal mess.

The World Intellectual Property Organisation (WIPO) should come up with Berne convention 3.0 that provides:

- increased protection for human authorship

- longer copyright terms for hoomans

- shorter copyright terms for bots (5 years)

- moral rights preventing 'globbing' by generative AIs

- royalty system for original authors

- derived computer-generated works pay

I'd like to see a blockchain ledger-esque system where human authors can claim authorship and they receive nano-royalties every time works are derived from it. The generative AI's can 'glob' it but they will need to pay their dues.

Importantly GAI's can't register their stuff in the hooman copyright chain.

I think the last thing we need is even longer copyright terms for humans.

"Based on the Office's understanding of the generative AI technologies currently available, users do not exercise ultimate creative control over how such systems interpret prompts and generate material."

As an artist and a musician I use various degrees of randomness in my work - from feedback and "noise" in a composition to abstract splatters in my paintings.

Plenty of art, writing, and music is the result of partially or completely random techniques, over which the artist has little to no control, and AI generated works are just another tool in the hands of a creator.

The copyright office understands neither art nor creativity.

I tend to agree, and I find myself doubtful that this will remain the status quo for very long. As soon as these things reach the point where they could realistically be used for something like a Disney movie, all of a sudden there will be a lot of money behind finding ways to make the output copyrightable.

It seems to me that there's a spectrum here. If, for example, I create an animated music video for a song (that I own) by using Stable Diffusion to generate the imagery, generate all the prompts that are fed to the model, cut and edit that imagery together via my own creative vision... Where does the line between my authorship and the model's end? I'm not allowed to claim any ownership of that output?

Then again... If someone enters a similar prompt, and gets similar output, should I have some claim to that?

It's a pretty tricky philosophical issue, honestly. The more I think about it, I think I'm ok with this as the general framework, at least for now, as, if nothing else, it may prevent powerful interests like Disney from gobbling up ownership of the whole generative AI space in its infancy

I think that's covered pretty clearly in this document? If you add creative vision on top of the generated output, then that creative vision is copyrightable.

Reply on news.ycombinator.com