The OpenAI board was right
garymarcus.substack.com
garymarcus.substack.com
But it also points to a wider problem in OpenAI's existence. They're built on theft. If you wrote something on the internet in the last 20 years, they've stolen it. They've stolen it, they're put it into their little magic box and now they want to sell it back to you. OpenAI is amazing technology, and once the equity of the company is redistributed back to the content owners who actually provided the value, it'll be great to see what they can do from there.
Jobs, at the very least, would have stolen competently.
OpenAI asked Johansson for her likeness; she refused. So they trained an AI on a person instructed to mimic her likeness, and then tweeted a mocking reference to the source they stole from [1]. Put aside your views on morality or legality. I’m damning them for being stupid.
[1] https://x.com/sama/status/1790075827666796666?ref_src=twsrc%...
And yeah, the techbro culture Sam steeped in results in remarkable stupidity!
Another "request" 2 days before release is a bullying, not stupidity.
Fuck around and find out, as they say. Johansson forced Disney to the negotiating table.
Attorney general of Delaware is Kathy Jennings <attorney.general@delaware.gov>. Mail your argument now. Do your civic duty.
1) Sam is a conniving weasel.
2) Sam has made many OpenAI employees rich.
3) The board knows that Sam is toxic and must be removed.
4) The board knows that removing Sam would mean the end of the company.
Exactly. Most people are simply hypocrites.
Sure. Now. I.e. only until they get sued.
it does have a connotation of advertisement to it, so I think it's worth examining. This isn't like complaining about how Apple doesn't sell food.
"Open" has a very specific meaning in Tech, and OpenAI is a tech company. It'd be the equivalent of Github swiching all of its repositories over to Mercurial tomorrow. They may have no obligation to support Git, but naming your company after such tech will have such expectation. You can definitely have an honest conversation about how using a word like "Open" for a tech company while switching to a closed source, publicly traded mentality is in fact a betrayal of the audiene who were inevitably attracted by the naming scheme.
https://futurism.com/researcher-openai-sex-drug-parties
https://www.themarysue.com/annie-altmans-abuse-allegations-a...
Erm... no. https://hachyderm.io/@inthehands/112006855076082650
> You might be surprised to learn that I actually think LLMs have the potential to be not only fun but genuinely useful. “Show me some bullshit that would be typical in this context” can be a genuinely helpful question to have answered, in code and in natural language — for brainstorming, for seeing common conventions in an unfamiliar context, for having something crappy to react to.
> Alas, that does not remotely resemble how people are pitching this technology.
He said "amazing technology" not "AGI that can replace humans". They are not the same.
> You might be surprised to learn that I actually think LLMs have the potential to be not only fun but genuinely useful. [...]
> I’ve heard from several students that LLMs have been really useful to them in that “where the !^%8 do I even start?!” phase of learning a new language, framework, or tool. Documentation frequently fails to share common idioms; discovering the right idiom in the current context is often difficult. And “What’s a pattern that fits here, never mind the correctness of the details?” is a great question for an LLM.
> Alas, the AI hype is around LLMs •replacing• thought, not •prompting• it.
This perfectly matches my experience too. If I know what needs to be done, then all LLMs waste my time on average. However when I'm trying to learn about some new topic then LLMs have been a good way to get me oriented in the right direction and get a really high level overview that allows me to get a jump start and start learning about the topic on my own.
They tend to shine a light on all the "unknown unknowns" (e.g. common terminology and concepts), but they're poor at converting "known unknowns" into real knowledge since they lack any real and deep understanding themselves.
Are LLMs useful? Yes. Do they live up to expectations set by their creators? Not even close.
https://www.reddit.com/r/singularity/comments/1cx1np4/voice_...
This whole controversy makes no sense and is obvious that the media is looking for something to attack openai with.
Apparently they selected the actress from auditions in early 2023, a while before the Johansson timeline.
Of course the OpenAI folk would likely have seen Her and may have selected someone sounding like that.
Because the most obvious thing for OpenAI do would be training the voice on Scarlett Johansson's available body of work, from movies, audiobooks, podcasts etc. Why bother hiring a voice actor?
Anyone can come up with DL networks these days.
If they deliberately used an actress that looks like her in an advert. A bit of makeup, a latex mask, digital touch up, full deep fake; at some point most people can no longer tell the difference.
Look at the comments, even in this thread, defending Sam. I thought this could be dealt with by the courts. But it clearly needs stronger medicine; there is simply too much hubris. Congress may need to act; let’s start with states’ AGs. (Narrow clarifications on copyright and likeness rights, to start. Prosecution from there. The low-hanging fruit is anyone in Illinois who was enrolled in WorldCoin.)
We, as a community, have been critical of crypto. AI is different—Altman is not. He is a proven liar, and this episode demonstrates the culture he has created at a multibillion-dollar firm that seems destined, in the long run, for—at best—novel conservatorship.
We have a wider problem. It isn’t “alignment.” It’s 19th-century fast-talking fraudsters clothing themselves in the turtlenecks of dead founders.
Of course we are. That doesn’t coerce anyone into a groupthink, and it doesn’t mean everyone is a part of it. I’m simply asking anyone who held one opinion to find harmonics with another.
I read HN 10x more than reddit just because of this fact.
Why do you equate acknowledging a community with group think?
HN promotes debate and individual opinions, so statements like in parent just have bad connotations to them.
Fair enough. I cited two we’s.
In the first—where “we, as a community, have been critical of crypto”—I juxtapose this community at large’s scepticism of crypto with its giving the benefit of doubt to people similar to crypto’s worst in AI.
In the second—where “we have a wider problem”—I intended to reference Silicon Valley at large. I did not intend to imply that Sam Altman is HN’s problem to solve. (Though I welcome the help, if you have sway in D.C. or your state capital. I am not sure I have the bandwidth to take on a political project of this magnitude.)
In consideration that Sam Altman was the former president of YCombinator, and Hacker News is financed by this company, it would be true hacker spirit (as in Hacker News) to subvertingly use Hacker News to "solve" the Sam Altman problem. :-D
Fantastic to see someone calling it out as BS.
End users who are not "developers"/"tech bros" who read HN cannot possibly be part of this imaginary "we".
That does not mean that everyone is on the same page, though. Probably the only thing everyone agrees on, is that this is a valuable personal resource.
If this was groupthink, I'd go somewhere else. BTDT. Got the T-shirt.
The world's a lot lonlier these days.
Credit where it is due: he corralled and inspired a difficult team. He oversaw the release of ChatGPT. His team wanted him back, because they believed he would make them rich. On that promise, he delivered.
Nobody has committed high crimes in this saga. But the lying has become endemic, from Altman to the culture. The situation screams for oversight, perhaps—most proximately—via consent decree.
This is why honesty is valued. If you look and sound squeaky clean, it's likely that you are honest and trustworthy. If you blatantly steal a famous actress' voice then lie about it, what else will you steal?
It’s behind the ‘you can’t con an honest man’ thing.
This is were you're very unfortunately missing the point that no, he doesn't, he is after something bigger.
my optimistic hypothesis is he really wants to control AGI because he believes he can make a more efficient use of it than other people. he might even be right and that's what scares me, because I don't trust his measures of efficiency.
I'd rather not let my pessimistic fantasies run wild here.
[0] https://www.cnbc.com/2023/03/24/openai-ceo-sam-altman-didnt-...
PG said the same- Never has he seen someone which such a capability to gain power.
He said Altman is extremely good at becoming powerful.[1] Did he say what you said also?
[1] https://www.newyorker.com/magazine/2016/10/10/sam-altmans-ma...
Of course it's money. He is a billionaire. What do you think it's about?
2) in a post-AGI world money may be an obsolete concept
Money helps with all those, but will not passively do that stuff.
we talk dollars all day long but we haven't quantified power nearly as well
lol
A lot of people want to take credit for it, but wasn't it started by a guy called Alec Radford ?
I've been working long enough, serious ageism kicks in if I say how long, but in this situation my decades of experience point at Sam as toxic, I can say all it takes is one idiot like Sam to destroy the work of a their entire organization. That "no for now" attitude destroys far far more than the issue they are pushing, it destroys basic trust.
But it's more nuanced for LLMs, because LLMs create derivative content, and we're going to have to decide how we think about and regulate what is essentially a new domain and method and angle on existing legislation. Until that happens, there will be friction, and given we live in these particular times, people will be outraged.
That said, using SJ's voice given she explicitly refused is unacceptable. It gets interesting if there really is a voice actor that sounds just like her, but now that openai ceased using that voice, the chances of seeing that play out in court are slimmer.
Google Search linked back to the original source. That was the use case: to find a place to go, and you went there. Way less scummy start than OpenAI.
ChatGPT does not help people find your content on your site. It takes your content and plays it back to people who might have been interested in your site, keeping them on its site. This is the opposite of search, the opposite of helping.
And robots.txt is a way of allowing/disallowing search indexing, not stealing all the content from the site. I agree that something like robots.txt would be useful, but consenting to search indexing is a long, long way from consenting to AI plagiarism.
Sure we could have. even if we're talking web 1.0 age, Congress passed a law as early as the early 00's for email, which is why every newsletter has to have a working unsubsribe link. so it's not impossible to do so.
regardless, consent is a concept older than the internet. Have an option of "can we use your data to X?" and a person says yes/no. It's that simple. we can talk about how much we cared 30 years ago, but to be frank that mistrust is all on these tech companies. They abused the "ask for forgiveness" mentality and proceeded to make the internet nearly impossible to browse without an ad blocker. Of course people won't trust the scorpoion one more time.
As an conrast, look at Steam. It pretty much does the same stuff on the inside, but it reinvests all that data back to the platorm to benefit users. So people mind much less (and will even go to war for) having a walled garden of their game library. Short sighted, but I understand where it comes from.
But why would anyone seek permission to use public data? Unless you've got Terms and Conditions on reading your website or you gatekeep it to registered users, it's public information, isn't it? Isn't public information what makes the web great? I just don't understand why people are upset about public data being used by AI (or literally anything else. Like open source, you can't choose who can use the information you're providing).
In the case being discussed here, it's obviously different, they used the voice of a particular person without their consent for profit. That's a totally separate discussion.
There's an unstated assumption that some authors of blog posts have: if I make my post sufficiently complex, other humans will be compelled to link to my post and not rip it off by just paraphrasing it or duplicating it when somebody has a question my post can answer.
Now with AIs this assumption no longer holds and people are miffed that their work won't lead to engagement with their material, and the followers, stars, acknowledgement, validation, etc. that comes with that?
Either that or a fundamental misunderstanding of natural vs. legal rights.
If people used the correct term for it, "lossy compression", then it would be clearer that yeah, definitely there's a line where systems like these are violating copyright and the only questions are:
1. where is the line that lossy compressions is violating copyright?
2. where are systems like chatgpt relative to that line?
I don't know that it's unreasonable to answer (1) with that even an extremely lossy compression can violate copyright. I mean, if I take your high-res 100MB photo, downsample it to something much smaller, losing even 99% of it, distributing that could still violate your copyright.
That seems like a bad example, I think you are probably free to even read the book out loud in its entirety to me.
Are you able to record yourself doing that and sell it as an audiobook?
What if you do that, but change one word on each page to a synonym of that word?
10% of words to synonyms?
10% of paragraphs rephrased?
Each chapter just summarized?
The first point that seems easier to agree on isn't really about the specific line, just a recognition that there is a point that such a system crosses where we can all agree that it is copying and that then the interesting thing is just about where the boundaries of the grey area are (i.e. where are the points on that line that we agree that it is and isn't copying, with some grey area between them where we disagree or can't decide).
In one case, you are doing it and society is fine with that because a human being has inherent limitations. In other case, a machine is doing it which has different sets of limitations, which gives it vastly different abilities. That is the fundamental difference.
This also played out in the streetview debate - someone standing in public areas taking pictures of surroundings? No problem! An automated machine being driven around by a megacorp on every single street? Big problem.
- a human will be an organic visitor that can be advertised to. A bot is useless
- A human can one day be hired for their skills. An AI will always be in control of some other corporate entity.
- volume and speed is a factor. It's the buffet metaphor, "all you can eat" only works as long as it's a reasonable amount for a human to eat in a meal. Meanwhile, a bot will in fact "eat it all" and everyone loses.
- Lastly, commercial value applies to humans and bots. Even as a human I cannot simply rehost an article on my own site, especially if I pretend I read it. I might get away with it if it's just some simple blog, but if I'm pointing to patreons and running ads, I'll be in just as much trouble as a bot.
> I really don't see a difference between this and someone reading a bunch of books/articles/blog posts/tech notes/etc etc and becoming a profficient writer themselves
tangential, but I should note that you in fact cannot just apply/implement everything you read. That's the entire reason or the copyright system. Always read the license or try to find a patent before doing anything commercially.
first of all it's not all public data. software licenses should already establish that just because something is on the internet doesn't mean it's free game.
>Unless you've got Terms and Conditions
The new york times did:
https://help.nytimes.com/hc/en-us/articles/115014893428-Term...
Even if you want to bring up an archive of the pre-lawsuit TOS, I'd be surprised if that mostly wasn't the same TOS for decades. OpenAI didn't care.
>Isn't public information what makes the web great?
no. Twitter is "public information" (not really, but I'll go with your informal definition here). If that's what "public information" becomes then maybe we should curate for quality instead of quantity.
Spam is also public information and I don't need to explain how that only makes the internet worse. and honestly, that's what AI will become if left unchecked.
> Like open source, you can't choose who can use the information you're providing
That's literally what software licenses are for. You can't stop people from ignoring your license, but breaking that license opens you wide open for lawsuits.
Part of that is that we've seen what Google has become as a result of that data harvesting. If even basic search engines are able to evolve into something as cancerous to the modern web as Google, then what sorts of monstrosities will these LLM-hosting corporations like OpenAI become? People of such a mindset are more vocal now because they believe it was a mistake to have not been as vocal then.
The other part is that Google is (typically) upfront about where its results originate. Most LLMs don't provide links to their source material, and most LLMs are prone to hallucinations and other wild yet confident inaccuracies.
So if you can't trust ChatGPT to respect users, and you can't trust ChatGPT to provide accurate results, then what can you trust ChatGPT to do?
> It gets interesting if there really is a voice actor that sounds just like her, but now that openai ceased using that voice, the chances of seeing that play out in court are slimmer.
It's common to pull things temporarily while lawyers pick through them with fine-toothed combs. While it doesn't sound like SJ's lawyers have shown an intent to sue yet, that seems like a highly probable outcome; if I was in either legal teams' shoes, I'd be pulling lines from SJ's movies and interviews and such and having the Sky model recite them to verify whether or not they're too similar - and OpenAI would be smart to restrict that ability to their own lawyers, even if they're innocent after all.
The courts are a lagging indicator.
This is the entire "move fast break things" play, it's legislative arbitrage. If you can successfully ignore the laws, build good will in the public, and brand yourself in an appealing way, then the legal system will warp around whatever you've done and help you pull the ladder up behind you.
On the other hand, if you break a bunch of stuff that people actually like and care about, politicians might feel like they need to do something about that. If you burn all your good will such that both political parties agree that you're a problem, that is when the play can go sideways.
The comments are overwhelmingly less defending Sam specifically and more pointing out that the voices don't really sound similar enough to definitively conclude theft from a specific person. Having heard a couple samples, I'm inclined to agree with them.
> We, as a community, have been critical of crypto.
"We" as a community have come to no such consensus. Speak for yourself.
In a way, it's similar to how cable TV started. Early cable relayed free terrestrial channels and charged a fee for it. The selling point was better picture quality in some areas. Cable operators did not pay broadcasters for the content they retransmitted.
It's not even remotely the same.
I can read a book, like it, create a screenplay with changes, and make a movie. It's different, but not completely different, and it is infringement.
I can even create an entirely new story, but using the same canon and characters, that too is infringement.
What people don't get is that you can't play games. TV lawyers are drama, and made up gibberish. Judges won't tolerate mental gymnastics, and attempts to explain away legalities rarely succeed.
But the parallel I was trying to draw is that they profited financially from other people's work without any compensation to the IP owners.
No. It never happened. Instead, legally mandated (in some jurisdictions, such as Canada), or negotiated deals allowed local cable TV providers to insert their own ads. This isn't "altering the broadcasts" without permission, which is 100% what we are discussing here. Instead, it's doing what is legally permitted.
Broadcast law is quite complex, and includes not altering anything at all without permission. It's simply not remotely the same.
You need to find another parallel here, I think. The logic behind cable television is rife with legislation, copyright law, a century of case law, and more. It is also co-mingled with broadcasting over the air, which is considered a public space, the airwaves being auctioned off (no, the internet isn't).
I don't understand the point you're trying to make about the significance of altering. Are you suggesting that if OpenAI simply served the content without altering it and didn't pay IP holders or obtain permission in any way, that would somehow be acceptable? This is what cable TV did before laws prohibited it.
As well, the TV show itself is not altered.
Was just thinking that sama managed to frame himself in opposition to the good guy Zuck.
This sounds absurd, but we're here, this is the timeline.
Anyone could beat Netflix or HBO or Disney or whoever if they just could take films and series and sell them without ever having paid licensing fees.
Anyone could beat Spotify or Apple Music if they just could take songs andsell them without ever having paid licensing fees.
etc etc you get the idea
What's saddest is they will probably impede OpenAI no more than they did Facebook.
The "Big Tech" companies trying to compete with OpenAI were built on theft, too.
They do not produce content. They collect data about people and usurp others' content for advertising purposes. This comes at the cost of privacy, "data breaches", teen mental health, journalism, perhaps even civil, democratic society, to name just a few sacrifices.
OpenAI is pushing those companies to be as bad as they can be. As we are seeing, they have no problem going there. It is "in their DNA". Always has been.
This forced showing of "true colors" is the effect of a potential competitor that cannot be merged and acquired.
People from all walks of life are noticing. It's a great time.
Altman is a master magician, he has even convinced the Vatican to write a document regarding the dangers of AI; in his path to reach bigger piles mountains of cash, he has thrown out all sense of ethics and integrity.
He has replaced entire board, solidified his position as the CEO, is actively attempting to silence any possible whistle blower or negative feedback internally through legal agreements, attempted regulatory capture to stop opposition, stolen other people's content to train his chat bot and of course now yet another scam, passing on someone else's voice as Scarlet.
It won't end here, he will keep on going and his followers will defend him, because they are truly convinced the man is a revolutionary.
I for one, fear for a world where openAI (open in lowercase is intentional) dominates with Sam at the helm.
I don't really expect any person or AI to be influenced by what I've written but if were I'm quite pleased. I wouldn't call it stealing.
Is everyone on this forum a thief because they learnt English and much culture without getting permission from those who came up with it?
Nothing was removed from anyone to train the model.
If I do a Christopher Walken impression, what have I taken away from the actor?
Nothing.
I stg it's so silly the contortions you anti-ai ppl go thru.
Clearly they wanted something reminiscend of the film "Her". Since Scarlett Johansson didn't want to do they voice they hired someone else. The voice they used is that of their voice actor, not Scarlett Johansson. She owns her own voice, but I don't think she has any rights to voices that merely sound like hers, when they're not being used to fraudently imply that she's involved.
So movies are really influential on culture, and I think there should be broad scope for "taking inspiration" as long as it stops short of plagiarism. Think how many works you've seen that are obviously heavily inspired by e.g. 2001: A Space Odyssey. I don't think the vibe of "I'm sorry Dave, I can't let you do that" as belonging to the voice actor Douglas Rain, or Kubrick or Clark. And that's what a good thing! Producing cultural works must feed the production of future works and I'd like to permit all but the most blantant ripoffs.
Sci-Fi Author: In my book I invented the Torment Nexus as a cautionary tale
Tech Company: At long last, we have created the Torment Nexus from classic sci-fi novel Don't Create The Torment Nexus
Ironically, his works are favorites of several leaders in the torment-nexus-creation space.
At this point, I have to assume that being out-of-touch is deliberate branding from OpenAI. Maybe to appeal to equally out-of-touch investors.
If this is the case, why did they pull it?
And why ask Scarlett again, just a few days before release?
This is all about public opinion, and they have communication managers and PR strategists. If they didn't train this voice on Scarlett Johansson, they would have had so much better of a response by not taking it offline and instead insisting it is not trained on her. The entire discourse right now would be very different. And you can't shift this back later, burned is burned.
The way I see it, there's 3 options:
· they did train it on Scarlett's voice
· they don't know if they trained it on Scarlett's voice
· their PR/communications people don't know what they're doing
I don't think #2 is any better than #1. If it's #3… idk, better companies have been ruined by poor PR.
For one, because it's obviously better to have the exact voice rather than an imitation. For two, because it avoids exactly this kind of sticky situation: even if you believe you have a legal and moral right to do something, it's always nice to have the question completely air-tight.
Also, remember that OpenAI isn't a single person, but a large organization with different people and viewpoints. Not unlikely somebody in the organization foresaw the potential for this kind of reaction, knew they couldn't convince the right people to pull the voice completely, but thought they could maybe convince people to reach out again.
I mean, suppose they had a Darth Vader-y voice for some reason and had approached James Earl Jones to do it, and Jones had said no. Is it really so terrible to have someone else try to imitate the Darth Vader character? Jones, along with Lucas, made Darth Vader a cultural phenomenon; I don't think Jones should have the right to dictate forevermore all references to that phenomenon. The same goes for Scarlett Johansson and the author / director of 'Her'.
This falls flat since the CEO tweeted "her" when the version featuring her voice clone was released. The CEO was bragging about this feature; the company was aligned behind it until the blowback went public.
https://www.npr.org/2024/05/20/1252495087/openai-pulls-ai-vo...
Hopefully you meant this as a joke, but this nightmare isn't so unrealistic! Put Sam's voice on top of gpt-5o and spin up fifty of them to start cold calling starlets, why not?
If it sounds too alike intentionally, it’s basically an impersonation of a character, which falls under copyright.
Someone else was sharing case law about this.
Regardless of whether they hired a different actress for the voice model, enough consumers would be convinced “ohhh it’s the same as ScarJo in Her!” which should point to there being an issue with ScarJo’s image rights and likeness, and although the law right now doesn’t explicitly cover AI voice modelling, as soon as this is tested in court it would likely set a precedent. Unless ScarJo lost that case, which seems unlikely.
However, the company hiring her might not have the right to use her voice with the intent purpose of passing that voice off as that of the other person.
She is completely unfettered in any roles that do not have that intent.
Please cite statute.
Generally speaking malicious intent only does not an illegal act make. Last I checked a voice is rightly not a trademarkable thing (it's a quality, not a definable entity).
Wow, damn, that's such a crazy ruling. How distinct is distinct enough? How famous is famous enough? I'm surprised Ford's lawyers didn't press the issue further up the courts.
Aren't they very explicitly not doing that? They're saying it isn't her.
They've lost any argument of coincidence, with just that one tweet…
That is, OpenAI was going to end up capitalising from Scarlett Johansson's voice, fame, and notoriety whether it is her exact voice or not, and without her permission.
It would be like me promoting a new AI movie featuring an AI generated character very similar in build and looks to Dwayne The Rock Johnson and promoting the movie by saying "Do you smell what we're cooking?" and then claiming it had nothing to do with The Rock and was just modelled on someone who happened to look like him.
Also, the value of a voice comes from (at least partly anyway), the huge amount of work, practice, skill, experience, from training that voice, experience in acting (and voice acting), and the recognition that resulted from all this experience and skill. Her voice wouldn't have "value" if she hadn't trained in acting, auditioned, honed her skills over many years. But OpenAI gets to use all that earned and worked for value for free?
This for me is similar to a writer honing his craft for a lifetime, 1000s of hours of work and labour, then someone training a model using his corpus to write plays and sell them without crediting him or paying him. It's trivial to make a model to imitate a writer, it's not trivial to become that writer to produce that work in the first place, so the writer needs to be both credited and compensated.
They paid the voice actor right? The accusation isn't that they trained on ScarJos voice, it's that they paid a voice actor that sounds like her.
The only thing in the list you gave is recognition that's different.
Johansson's accusation did not specify the method. OpenAI claimed they hired another actor and the similarity was unintentional. The 2nd claim seems unlikely. Some disbelieve both claims.
Algorithmically? How much of a similarity score is too much? How much vocabulary must you compare? Suppose two voices sound entirely identical, but one voice has Canadian aboot. Is that sufficient difference?
You can maybe see why the courts wouldn't want to touch this with a 1 kilometer pole
https://en.wikipedia.org/wiki/List_of_songs_subject_to_plagi...
Given the courts regularly deal with murder via decapitation or incest, I’m sure “The Case of the Attractive Actress’ Voice” would be a nice day out for them.
Suppose you turn down a job offer as a programmer. Can you be pissed at me if I offer the job to someone whose code looks like yours based on scraping GitHub?
Basically your voice is part of your likeness.
"Impersonation of a voice, or similarly distinctive feature, must be granted permission by the original artist for a public impersonation, even for copyrighted materials."
Did you even look at the Ford commercial? https://youtu.be/hxShNrpdVRs
Having someone sing in the exact same style as another singer is totally different from what OpenAI did with their voice AI (having a female actor speak in a flirty tone).
It makes sense with music but you're setting a really dangerous precedent if you can't even hire a voice actor who sounds similar for speaking.
Of course she can. She just can't fool consumers into thinking that she is SJ.
The problem here is not that the voice sounds identical to SJ, it's that the consumers are led to believe they that is SJ's voice.
That’s misleading and blatantly illegal.
Someone stole from a bank and your argument is “are you saying it’s illegal withdraw money from a bank with vigor?”
Because they could be lying, they could have used a different actress and also mixed some Scarlet voice into the mix.
https://x.com/sama/status/1790075827666796666?lang=en
"her" -Sam Altman
She might!
https://mcpherson-llp.com/articles/voice-misappropriation-in...
IANAL but it's actually not that simple. There are laws and precedents around soundalikes.
I think the cultural defense for this is really lame. Was there really cultural value in evoking "Her"? Because outside of the narcissistic male nerd culture of laughing at your own jokes, and fixating on your own cleverness, there doesn't seem to be. No one outside of this bubble actually thinks it's super cool and culturally valuable for tech companies to attempt reinvent - seemingly without a trace of irony or self-awareness - the cautionary tales and trappings of dystopian sci-fi.
From what I have seen, the demos struck the average person as cringey and left women in particular with a deep sense of revulsion.
Especially if you then post references bragging that your thing either is, or is basically the same as, that other thing.
I.e. they asked for permission (which they presumably felt they needed), then launched not only without it, but without a response either way.
> As a result of their actions, I was forced to hire legal counsel, who wrote two letters to Mr. Altman and OpenAI, setting out what they had done and asking them to detail the exact process by which they created the ‘Sky’ voice. Consequently, OpenAI reluctantly agreed to take down the ‘Sky’ voice.
Strongly suggests OpenAI had something to worry about regarding their own process.
This way, legally you're in the clear, because you didn't use the voice of X. Not on any occasion did you use the voice of X. You even paid a voicer to make your recordings, and it's their voice.
Morally, it can be - and probably is - a different matter.
You can argue about these morality of that, but I think it’s reasonably true, practically.
Thus, it got taken down.
No, that would be either cortana or Majel Barrett Roddenberry.
Would the everywoman on the street be able to recognise Majel over Scarlett? I don’t believe so.
(i.e. I'm not sure if I would jump to "that's HAL!" if it weren't in disobeyance and vaguely threatening…)
Citation needed.
Even when VAs are recast for reasons beyond control (death, unavailability, aging child voice, health, etc.), they are only similar, not imitations. Happened in a lot of anime.
First is the belief that the essence of a person can be reduced to quantified metrics as inherited and objective as height or, in this case, the shape of one's vocal cords and the resultant pitch of their voice. Second is using a glorified function approximator as an arbiter. The positive outcome for OpenAI would be a classifier able to achieve high accuracy in distinguishing between the original and impersonated voice as evidence that the voices are sufficiently distinct.
In the event of an underperforming classifier, what is preventing OpenAI from claiming that the current technology is simply not adequate? Under what grounds could one refute the claim that a future, more advanced system, very probably trained with more appropriated data, would be a better classifier and favor OpenAI?
This is not the point, as stated by Ms Johansson's statement, her own friends could not tell the difference which is the point. If a reasonable person cannot tell the difference, there is none. In combination with seeking a prior agreement to use her voice this sounds quite dubious at best, malicious at worst.
There's a nice comparison here. https://www.reddit.com/r/singularity/comments/1cx1np4/voice_...
I bet OpenAI took it down for the free PR, when life hands you lemons...
ChatGPT using Sky voice (not 4o - original sept. release): https://youtu.be/JmxjluHaePw?t=129
Samantha from "Her" (voiced by ScarJo): https://youtu.be/GV01B5kVsC0?t=134
Rashida Jones Talking about herself https://youtu.be/iP-sK9uAKkM
See what happened in Back to the Future Part 2: https://en.m.wikipedia.org/wiki/Back_to_the_Future_Part_II#R...
Plus this opens the door for all kinds of things. Say tomorrow some one makes a AI that sings. And they make any lyrics, and make a AI sing it, say in a voice that's very close to their favorite singer. Would that qualify as illegal?
Would be interesting to see how that legal side of things in this regards unfolds.
With purely synthetic voices, it’s gray area just like with most of the other genai.
Besides, how does one prove a recording is of a particular person's voice? The AI might be generating voice similar to that of Scarlett Johnson, but its still not her voice. Similar != Same.
The fact that the future of AI is in the hands of one of these individuals is an utter disaster.
It's a slippery slope.
The problem here is that they wanted Scarlett Johansson and imitated her when she said no.
There is also a big market for "celebrity sound alikes" in voice acting.
The case here is that there's a voice that sounds like Scarlett Johansson, but OpenAI are saying that its isn't her.
The problem for them is that they've already admitted that they really wanted it to be her. They should have consulted a Hollywood lawyer for this one.
Well, here's Sam Altman quite literally implying that it is her: https://x.com/sama/status/1790075827666796666
The problem for OpenAI is probably that this is the first time that someone with a bit of standing caught them with their fingers in the cookie jar and not only have the balls to tell them no, but also have established rules to go after them.
Satire has protections that other fields lack.
If they would have never mentioned Scarlett or asked her, it might be different situation.
They wanted someone that would be able to perform a specific character (the AI in the movie "her"). They were looking for the right person for the fit, they got their eyes on Scarlett (of source, she played the character, originally), and when she refused they found someone else who would be able to deliver a close performance. It happens all the time.
To me, the "sky" voice is way closer to Rashida Jones rather than Scarlett Johansson: https://www.youtube.com/watch?v=dJmRaTogqrM&t=370s
- There is a finite number of distinct human voices.
- The # of desirable human voices (D) < The # of distinct human voices (A)
Where D := A - (All unwanted / undesirable human voices)
- Two voices that sound alike are treated as if they are alike
The fact that people say that it sounds like Scarlett Johansson means that there IS a region of multi-dimensional vector space that is claimed to be "owned" by the actress and that actress alone.
I HATE the end conclusion that came resulting from this mob spectacle: That people with voices that sound like hers have effectively no agency with what they can do with their own voices. That they are eternally marked as "that person that sounds like Scarlett Johansson", and that whatever they do with their own voice (including consent to be used for ML training) is met with the ire of the online mob.
Whoever voiced Sky didn't lose their agency over their own voice. The issue at hand is Sama not taking No for an answer.
Yes, OpenAI did not explicitly say it's her voice, but Altman tweeting "her", it being called "Sky", them asking her for permission twice, they definitely wanted this connection (and her real voice), IMO, and like so many tech companies, they just did not accept "no".
They could have used the same voice, called it "Moon" or whatever and stopped Altman from tweeting, and it probably would have been fine.
In this context, I understand the result. Without this context, I would not, and would agree with your points.
[1] Or maybe just used her voice samples anyway in some form, that's hard to proof / disproof without an external audit - depends on whether or not you give OpenAI the benefit of the doubt
However, the principle still stands.
Proving intent is extremely hard and I always wonder how anyone that has enough power to pay attention let's their intent be documented. At least for Sam it sure seems to be due to a massive amount if arrogance.
If Sam was less vocal about his love for her, we wouldn’t be out here thinking that OpenAI is trying to ride off the back of the movie.
Context matters. Adobe cannot defend its logo in all contexts. The laws make this clear. You are ignoring the context of Scarlet Johansen portraying an AI character in a movie then her likeness being used in another AI character without her consent. Had she never starred in Her, had she never been approached by OpenAI, this would not be a story.
https://www.engadget.com/adobe-threatens-to-sue-nintendo-emu...
Think she is a prominent actress that's why this news is all over the internet but every individual has the right to their voice and face shouldn't be cloned, for them it's not concerning on internet but definitely in their family, relatives and society
However we solve for that will tell us if and how we can motivate people to make all the tradeoffs needed to stand out and bring new unique art into the world.
I am hoping we land somewhere in between the extremes of "no copyright use" and "what's a copyright" and create a royalty system that can trace your content's influence on the weights used at inference and cut you in according to your contribution.
However!
With this mentality, the mentioned paid voice actresses voice is not hers anymore because another one's voice is similar and it's used in a movie? So she can't work and voice any other work without Scarlet Johansson's permission?
It seems they never considered it possible that someone would say no to them and would protect their IP against them, because who wouldn’t want to be part of their glorious mission, right?
As long as you don't commit crimes, this strategy will always work out. Hell even if you do commit crimes, as long as you have willing "fixers", you got to piss off a lot of powerful people to face justice. Just look at the 45th - that guy has a history of decades of stiffing contractors, lenders and employees (including those who actually went behind bars for him), but it took a pretty flimsy constructed case to finally charge him with a crime and I'm not certain that the case is clear enough for a conviction.
> I probably don’t have to tell you, but that’s complete bullshit. And stupid, obviously refuted bullshit at that.
This article is quoting the linked Tom's Guide article but nowhere in the original OpenAI statement do they claim that their voice's resemblance to Scarlett Johansen was "purely a coincidence". It's kind of important to actually quote the original source before you call it out as "complete bullshit".
Original OpenAI statement: https://openai.com/index/how-the-voices-for-chatgpt-were-cho...
Not the exact wording, but close.
If this is OK (and of course it is, no one has to accept any offer), how is OpenAI working with someone with a similar voice tone not OK?
If Sam used someone's voice without consent, that's illegal and immoral, but I see nothing wrong with Sam paying someone with a similar voice and using that voice instead.
The “her” tweet makes it exceptionally clear that it was always intended to be Johanssen’s likeness, not some other voice actor’s that happens to sound similar; OpenAI is just betting they can get away with it.
It seems like OpenAI were also betting on Johanssen not knowing her rights _de jure_in that area, which it appears they were quite wrong about. (And even if she didn’t, she would have quickly learned of them in this hyper-connected world, making the bet utterly incompetent…)
What's wrong with using a similar voice to the one in your vision of the demo/service, as it's the closest voice that you can (legally) use it?
Like, for example, voicing an AI chatbot when that celebrity, one of the most highly paid in the world, has previously starred in a famous movie as an AI chatbot. And especially when the makers of the chatbot have repeatedly publicly referenced the movie as inspiration in creating the chatbot and have publicly pointed out the similarity between their chatbot and the movie. And especially when the makers of the chatbot have approached that celebrity to ask to use her persona and been denied.
Because if you do that you might just get your arse sued off.
Intent matters a lot in a situation like this and OpenAI very clearly documented their intent for every me to point and laugh at. Even in the hypothetical case I give above where they don’t try to hire Johanssen and just get someone similar-sounding, they could still be liable for appropriating Johanssen’s voice if they were coaching the voice actor to specifically sound like Johanssen rather than do her own performance.
You’re looking for a technicality but this stuff relies on human judgment for a reason.
As long as it's not literally her voice (or a model trained directly on her voice) they should have rights to do anything they want.
If a similar voice was used because it was _reminiscent of_ “Her” but was still a unique performance, then it probably wouldn’t be a problem. Where there’s a real problem is if it’s an intentional mimicry of a specific person’s specific vocal performance.
And Johanssen can show OpenAI had such intent in court pretty easily, between their attempts to get her consent to use her voice (which she denied) as well as their CEO’s “her” tweet promoting the voice.
That person's (Scarlet) voice just happens to be a great match for the intended tone and since they refused, they went to the next candidate.
Do you really think that everything exists in a a vacuum and must be decided from first principles, giving the maximum benefit of the doubt to one party over another? There is extensive case law on exactly this situation, and OpenAI has run afoul of it.
The legal issue is that there are specific, established laws based on ideas like the above, known as "right to publicity" laws.
I feel bad for the voice actor OpenAI hired. She was probably really excited about this and then SJ had to make it weird and claim she was ripping her off
It wasn’t SJ’s natural voice in Her, it was a performance that is nothing like her regular speaking. The idea that she can claim giggly/pitchy as only her domain, or that she owns the rights to the character in the movie, feels really silly to me
But after a few rounds, no one would help him anymore and he would loose all the games.
"Microsoft gets a lot of heat for not leaving enough on the table for others. My mother, who’s English, and quite a character, once said of Bill Gates, “He sounds like someone who would come to your house for dinner and say, ‘Thank you. I think I’ll have all the mashed potatoes.’”"
Replace Gates with Altman.
source: https://daringfireball.net/2013/05/all_the_mashed_potatoes (the original post is gone from the O'Reilly website)
I originally got this sentiment partly from this: https://news.ycombinator.com/item?id=38378216
... because not pulling the trigger clearly isn't an option.
The unauthorized use or reproduction of humanity's work.
There is a lot of nonsense and drama about all of this, but the voice is clearly not Scarlet, even though they were clearly trying to get someone to sound like her. So much drama over nothing.
They approached Ms. Johansson, she said no, and they went ahead and did it anyways.
They hired a voice actor who had a similar voice profile to her, a similar voice to what they were looking for. If you just hear the voice they used, it sounds very similar, but if you actually play them side by side, you can hear they are very clearly different.
How is that different from anything else? You ask someone to do something you want them to do, they say no, so you then go and find someone else that will do what you need which is what they did.
And Sam Altman's lower case single word tweet "her" clinches it. Clearly he was not just announcing his new pronoun.
Having something similar to someone else does not equal theft. What is wrong with people? Why are they acting so weird about this.
What about the voice actor that did this? Is her voice 'theft'?
https://x.com/sama/status/1790075827666796666?lang=en
And you already admitted "they were clearly trying to get someone to sound like her". You pre-lost your own argument yourself, before I even replied, with your own words.
Why are you acting so weird about this, like there's something wrong with you, and then trying to project it onto other people?
Trying to emulate Trump's debating techniques and psychological disorders doesn't work any more, nor look very good on you.
It doesn't' change anything, they can make it seem similar to her if they want by hiring a voice actor with similar characteristics, the voice actor is paid for there work and they get something that is great marketing. There is no theft here. This is a completely normal situation, your massive overreaction is not normal.
If they stole her voice outright, there is a problem, but it is extremely unlikely they did. So then there is no issue. It's common sense, i'm not sure what's wrong with you that you are trying to bring politics into it.
It is absolutely not clear if that's the case; and there is caselaw that suggests that they absolutely _can't_:
https://mcpherson-llp.com/articles/voice-misappropriation-in...
Should the voice actor who sounds a bit like Scarlet just quit working?
Since OpenAI asked her for permission two days before releasing it, obviously THEY thought it sounded like her voice, and that they needed to ask for permission. Making public claims implying that she's involved with the project like Sam Altman's "her" tweet, AFTER asking for permission and receiving two "NO"s, proves malice and dishonesty.
Sam Altman's tweet is going to be Exhibit 1 in the lawsuit he loses.
https://x.com/sama/status/1790075827666796666?lang=en
It makes me wonder how many "NO"s Sam Altman's sister Annie told him, that he also ignored. Maybe SHE deserves HER voice to be heard.
https://x.com/anniealtman108/status/1459696444802142213
Annie Altman’s Abuse Allegations Against OpenAI’s Now-Fired CEO Sam Altman Highlight the Need to Prioritize Humanity Over Tech:
https://www.themarysue.com/annie-altmans-abuse-allegations-a...
>What are Annie Altman’s allegations against her brother Sam Altman?
>Annie Altman alleged in a November 2021 Twitter thread that she “experienced sexual, physical, emotional, verbal, and financial abuse from my biological siblings, mostly Sam Altman and some from Jack Altman.” She returned to those allegations in 2023, seeming to call out Sam. “I’m not four years old with a 13 year old “brother” climbing into my bed non-consensually anymore,” Ms. Altman’s account, @phuckfilosophy, tweeted in March 2023. “(You’re welcome for helping you figure out your sexuality.)” [Sam Altman is openly gay, and there are nine years between Sam and Annie.] “I’ve finally accepted that you’ve always been and always will be more scared of me than I’ve been of you.”
When Kinect (remember that lol) was being released people compared it to minority report on stage. Does that mean Tom Cruise should have sued Microsoft?
Step 2: use control vectors to make it slightly different
/s
Repeat after me: Sam Altman is powerless if we all ignore him. Sam Altman is nothing without his people. He can't do anything himself.
Questionable ethically perhaps but OpenAI doesn’t seem to care about that
But of course, all this is not about only one man. Corporatocracy will prevail, if not actively confronted.
The voice, all names, characters, and incidents portrayed in this AI are fictitious. No identification with actual persons (living or deceased), places, buildings, and products is intended or should be inferred.
Advance digital intelligence in a way that benefits humanity, without the need to generate financial return.
That bit about financial return has since been removed.
Until now no one could point to a human story of what that meant, instead pointing to the shift from a 501(c)(3) to a for profit model which people don’t understand so it just doesn’t resonate.
But now, we have a human story of someone that makes a living on their likeness, the CEO approaching them for licensing rights, then allegedly just training the AI on their likeness and stealing it for financial gain without compensation when the CEO couldn’t get a deal done.
AI is dystopian to start with, which is why the initial model was created as a non-profit to benefit everyone, but now we have a human story validating the danger and dystopian beliefs surrounding AI.
On top of everything the Founder was personally involved in making these decisions and even Tweeted “her” as a sick joke referencing the actress’ movie where she voiced an AI.