Claude for Creative Work
anthropic.com
anthropic.com
> Notice: This announcement is causing a lot of feedback. We are actively evaluating it.
Presumably a lot of Blender users work in roles that feel threatened by AI being used for computer graphics work.
Lots of negative replies on Blursky here: https://bsky.app/profile/blender.org/post/3mkkuyq3ijs2q
This feels like the proper way to have AI act as a tool to make artist's jobs easier without taking away their creativity?
Edit: I guess they might want absolutely no AI of any sort in their tools (which seems like a strange line to draw), or is it about the data it's been trained on?
Even if you can see how individual circumstances could be beneficial to your workflow, it's a general direction I think many take issue with quite fairly.
But what's the plan, then? Prevent any third party from downloading Blender and integrate it in any way with an agent?
I fully expect things to march on anyway. I have no idea how it plays out for creative industries, I am still thinking and observing in that regard.
Without a constant stream of stolen training data, the "AI" piracy bleed-through and isomorphic plagiarism business model is unsustainable.
We look forward to liquidating the GPU data-centers at a heavy discount. =3
I'm skeptical of this line of reasoning. Major content providers have no problem with copyright, even when content is completely produced by anonymous contributors. Is this supposed to become an issue when you eliminate some anonymous contributors?
Besides getting sued for piracy, settling out-of-court with Disney, and or externalizing DMCA/RIAA take-down liabilities on users.
A human may transfer rights or "license" to another party in many circumstances, but may not re-sell a codified Coca-Cola logo trademark out of convenience.
All levels of the US courts concluded an "AI" can't transfer nor actually create content rights. Most WIPO members also seemed to follow the same consensus.
https://www.bbc.com/future/article/20260414-the-monkey-selfi...
There was a similar issue with folks selling marginally pitch-shifted audio assets on the Unity and Web stores. Note, they didn't have the original legal right to license this content, and customers would get their content flagged eventually.
Some kids are cheeky pirating Sony and BBC libraries... exploiting peoples assumption buying an old CD set somehow magically gives the holder broadcast or game distribution rights.
Keep being skeptical, as it will keep you in business. =3
The "how" it happens part is just legally irrelevant "[piracy] with extra steps", but if you are interested in details see below. =3
https://www.youtube.com/watch?v=YhgYMH6n004
https://www.cbsnews.com/news/taylor-swift-ai-voice-likeness-...
Here is a simplified explanation of how vector search is done in many models:
https://www.youtube.com/watch?v=YDdKiQNw80c
And a more detailed toy implementation to learn how to build your own:
Your comment makes no sense. The whole concept of "piracy" is meaningless when applied to LLMs, unless you go way out of your way to prompt models to output specific works verbatim.
Also, you do not "pirate" Harry Potter if you prompt a model to generate a story that directly or indirectly involves Harry Potter in any way. Like always. You can argue trademark violations or copyright violations if someone tries to use said work for commercial purposes, but LLMs are orthogonal concepts.
Just because Photoshop allows you to hack together variants of the coca-cola logo that does not mean Adobe is liable for trademarks or copyright violations.
LLM bot poisoning discourse is against YC site usage policy.
>you do not "pirate" Harry Potter
True, but firms broke the law acquiring the content, and copyright violation occurs if the output bears similarity to existing works. The cited lawyers analysis explains how violating likeness applies to everyone now regardless of notoriety.
Again, the black-box argument for washing ownership rights is a fallacy, and the links covers how LLM are built. There have already been several dozen precedent cases showing LLM output is mostly weakly obfuscated intellectual property.
Notably, the training data also includes other LLM users markdown data.
>Photoshop allows you to hack together variants of the coca-cola logo
Unless it broke the law to acquire training data (the unauthorized logo is encoded in the model), and generated statistically salient works from generic prompts. For example, "Name a cartoon mouse" will usually output Disney Mickey Mouse trademarks, rather than Mighty Mouse.
LLM are quite good at content search, but are a confirmed liability. =3
I don't know what that's supposed to mean, but I'm afraid it sounds something that involves tinfoil-based head gear.
> True, but firms broke the law acquiring the content, and copyright violation occurs if the output bears similarity to existing works.
Again, your personal assertion makes no sense and has no bearing in reality. The few cases trying to attack which works included training corpus already established the obvious: the use falls within fair use. To question this fact you would first need to assert that you could violate copyright by glancing at a book the wrong way.
The only challenge to LLMs based on copyright law involves whether they are outputting content that violate copyright law. Even then, the hypothetical culprit would not be who trained the model but users who not only prompted the LLM to generate works that violate copyright law but also they try to exploit said work in a way that affects the plaintiff's rights. I'm talking about things like some random person prompting a model to output a book about a wizard called Barry Potter, and publishing it somewhere. Those hypothetical cases involve model users and copyright holders, not LLMs.
> Unless it broke the law to acquire training data (the unauthorized logo is encoded in the model),
There is no such thing, even in jurisdictions with draconian copyright laws such as the US. I recommend you spend a few minutes googling for cases that were in the news already.
Sock-puppet accounts may be banned for AstroTurf or slop.
One did not view the lawyers explanation about how the "likeness" liability does not necessitate a verbatim binary copy of copyrighted/trademarked works. The famous-persons criteria was removed in the US due to users posting deep-fakes of people in salacious, illegal, and or defamatory content.
The weak obfuscation/compaction of pirated and plagiarized content is provable in many "AI" models, and papers were posted by other YC users detailing how one may verify this yourself by intentionally outputting the original training data:
https://arxiv.org/abs/2510.15511
>There is no such thing, even in jurisdictions with draconian copyright laws such as the US.
It is actually very common to charge people engaged in piracy of IP. Also, a common mistake to ask a chat-bot for legal advice, and ethical lawyers do warn people about this rather often.
https://en.wikipedia.org/wiki/Theft_of_services
>I don't know what that's supposed to mean
The instant people pirate content in a commercial setting, the clock starts ticking on legal peril. But there are simpler explanations of what models "do" available:
'"Generative AI" is not what you think it is' (Acerola)
https://www.youtube.com/watch?v=ERiXDhLHxmo
ymmv... Best of luck =3
A lot of artists who would love to be able to direct their professional software in natural language have to reconcile that with how this technology came to be and what the aims are of the company now delivering it to them.
Software tends to be a "living" project, so just vibe coding with 0 software knowledge is not yet fully sustainable for maintaining a project. But with art, the AI just spits out a completed image.
The generated images compete directly with the people the data was sourced from, and there have also been many cases of abuse, eg people using AI to impersonate a popular artist and selling comissions under that artist's name.
The copyright situation for generated imagery is also tricky, so people pretending to be artists only to be sharing work that isn't copyrightable can cause a ton of trouble and financial loss for customers.
Most of these issues don't apply to software in the same way. That's why I was surprised by the backlash to this as it's just touching the software side, I don't see this as threatening artist's work.
When I was dabbling in image generation (~StyleGAN2 era), my vision for image generation models was as a support tool for artists (back then I was generating small character thumbnails to help me brainstorm ideas for drawing), believing that people valued art for the human effort. Even then I would have considered what Anthropic are trying to do here as the preferable way to use AI in art workflows.
They are protesting against natural technology development. To me it looks similar to taxi drivers protesting against Uber (protecting their right to scam tourists).
Did drawing artists protest against photography? Do celebrities protest against photographers selling their photos taken by them in public places?
They are right to be afraid though. What's really happening here most probably is Anthorpic buys rights to collect user trajectory data. In order to replace Blender users later.
Oh, I wish it was limited to lifetime.
USA is currently lifetime + 70 years, and work for hire is 95 years from creation.
I would think the same goes for codebases too. On a personal note, I wrote a CMS in Elixir from scratch way before even AI was a thing. It uses a lot of proprietary flows to make it scale, helping it serve millions of requests efficiently. I certainly did not give OpenAI nor Microsoft permission to steal my code. And yet they did. Is that not theft of my Intellectual Property?
Neither is fair use, and neither is copyright infringement. But learning most definitely is not theft.
There is no such thing as "stealing" copyrighted work. Either you have unauthorized access and/or distribution, or you don't.
Unauthorized access to copyrighted work is perfectly legal in a big chunk of the world, including western Europe. Read up on the french tradition of copyright law, particularly the provisions for personal use.
This brings us to how "people have gone to prison for simply pirating DVDs and CDs". The bulk of the cases were focused on mass commercial distribution of verbatim copies of third-party content. I'm talking about DVD-burning factories.
Maybe true in places with different cultural values like China or India.
However, piracy differs from the theft-of-service data scrapers use while ignoring EULA, site usage terms, and robot exclusion standards.
https://en.wikipedia.org/wiki/Theft_of_services
Getting emotionally invested in "AI" fantasy marketing is irrational. Have a wonderful day =3
No, this is a core trait of the whole concept of copyright.
Copyright is a legal tool to allow authors to claim the exclusive right to monetize their work. But from it's inception this same legal tool is designed to ensure the public has the right to access said copyrighted works without authorization, including but not limited to the right to the unauthorized access for personal use and how public domain is extended to all works.
This notion originates from France's copyright law, from which all copyright laws in the world directly or indirectly comes from. We are talking about centuries of legal history.
We are in the age of "Napster" for nonsense, and "free" stuff other people made is certainly a crowd-pleaser. =3
For example, you could least feel that the world is large enough to have people with other needs, drives and ownership levels of their work.
You could also consider that this is not an even trade; artists had all their works ingested and didn’t get a commensurate stake in openAI.
You can consider that you had a choice to share when you contributed to open source. Then imagine how a counter culture artist, who despises corporate culture, must feel to have their work consumed by another rapacious tech entity.
Or you can be the filmmaker whose clients are now showing up with entire ad clips, and then decide they would rather not spend the money on CGI to complete the video - essentially demolishing work overnight.
This isn’t to say that there are not artists who are excited by this, or artist who are happy to have their art ingested. Just that the way you phrased your question evoked this answer.
Makes me think that there's some room in the model lineup for one that doesn't do as well on benchmarks, but is trained on "ethically sourced" data (though they'd need to somehow prove that they aren't "accidentally" including other data).
It almost totally automated vast swaths of texture generation by creating algorithmic systems that technical artists could use to create textures.
Want a brick texture? Sure, you connect some nodes and set parameters and you have great looking bricks. Want the mortar to be a little more widely spaced? Done. Want some moss on the brick? Want some chipping on the brick? Want some color variation? Done, done, done.
It probably reduced the amount of time to iterate textures by more than 100x.
Now, talented technical artists make OK money because they are good at using these tools. Photoshop jockies are gone.
LLM manipulation of Blender will be interesting but it's very, very challenging to see the path of something like Claude having nearly as big of an impact. It'll be helpful to automate some common tasks, to build internal tooling. But Allegorithmic single handedly changed the way 3D games look, because you could be so much more ambitious.
You didn't really hear about it, though, because it wasn't part of the cultural zeitgeist.
They are conscious of preventing momentum in a bad direction.
If they don't fight it hyper hard, a huge fraction of them will be out of a job instantly.
Given how much software and other AI/computer vision improvements 3D content often relies on, it's weird to decide that the algorithm itself is unallowable.
"If you ignore their biggest, their primary, concern, their other concerns seem almost trivial".
So, what I meant when I said '... otherwise ...' wasn't trying to dismiss the data sourcing concern, but more like "I understand if the data sourcing is the concern, but you (make3) seem to be saying it's about the use of AI in general (ie even if, hypothetically, an ethically sourced training dataset was used for a model), which feels like a weird restriction to me". That was when I added the edit to my initial post.
AI is seen as an oppressor and a threat, and AI providers are seen as oppressors. It's understandable that people don't want to collaborate with their oppressors, either direct or by association. If you were a Jew, would you buy shoes from the Nazis just because you were individually safe from them at that moment? Or would you if you were of a minority they hadn't started exterminating yet? Or if they were not exactly the Nazis killing your people but some affiliated group?
This sounds extreme until you realize they are under threat of losing their likelihood for good.
They are right to not accept your inevitability point without a fight, this is a human thing that can be fought, revolutions have happened, and will continue to happen.
I don't necessarily agree with this but I do understand it.
You can find AI useful and still be against its introduction into your field for entirely understandable reasons.
Unfortunately this does create uphill friction for any good-intentioned people trying to use AI to improve art by empowering people to take on more ambitious projects. (This is a general statement and not related to the case of Anthropic. Of course Anthropic here is just trying to sell their product, which is a fair thing to do in isolation, but I also understand the opposition to it on the grounds of its downstream effects.)
For all of us, acceptable use is when I use AI to do my job. Unacceptable use is when you use AI to do my job.
AI removes all these hurdles and directly presents you with the end problem - communication. Artists hate that because most artists don’t have anything to communicate. These people deserve to be automated away. I don’t wanna see more derivative shit. Artists who have something special to communicate won’t feel threatened by AI but feel more freedom.
Which is why 99.9% of AI art is worthless. There's literally nothing personal or interesting about getting grok to fart out some picture you thought about while sitting on the toilet in the morning.
AI art will never be good without actual artists embracing the medium.
Even myself, while I am currently extremely empowered by these tools... I could see my role (Founder/PM/builder) disappearing in the next couple years.
I respect you a lot, so if you have a moment, I would really like to get talked down from my take.
I still had b2b product ideas, but was never able to raise funding. I built an MVP in 2 weeks, got a co-founder to pay for my beans and rice for the year spent building, getting feedback, and refining. We now have paying users.
Which it seems like they can just choose not to use?
Maybe the blowback has more to do with gatekeeping.
I doubt the current state shows the end of their ambitions.
It'll be way easier to understand for developers when it starts happening in earnest to our profession, which is coming soon.
It already is here to some extent, but so far mostly on the junior end so it hasn't been impacting many people who are already established in an industry that has provided relatively easy stable livelihoods for the past 30+ years, but soon won't.
The developers are literally on the bleeding edge here, it might be the most developed of the AI use cases right now. The most advanced tooling for LLMs revolves around SWE work, there are multiple prolific benchmarks that the labs are actively targeting in this area, and new ones are being built, whole product categories being spawned, software companies bleeding money for tokens.
It's the other professions that are to follow once the training data is in place to go reach for their livelihoods. SWEs got the early taste of what is coming. And the blender news is exactly that.
To the surprise of no one.
I understand being unhappy about something but people gotta relax.
---
Just like AI image slop and AI book slop prove though, I highly doubt whatever Claude and Blender are cooking up will ever come close to taking a prompt like
> render a scene of a corgi sitting on a chair looking out of a window at 3 cats playing with the corgi's favorite toy.
and turning that into anything useful.
It is a massive SDK though (thousands of functions; feel free to poke around with it; Affinity is free) and so it really shows the ability of LLMs to effectively work across long-horizon tasks massive context windows.
Personally, really interested in Blender though. I'm working on a game as a hobby/side project and I'm very much a newbie / often struggle with learning and using Blender.
There are so many ways these integrations help humans & human creatives; your job and role shouldn't be about how skilled you are with navigating/using a tool, or if you're technically savvy to code scripts to improve your workflow.
Turns out it is possible, one just has to have the script check to see if each level of a given index entry exists or no, then if it does not yet exist, create it before making the next lower level by adding that sub-entry to the one above it.
An LLM is only going to code what it has documented as possible/working and may not be able to do what needs to be done.
I think a big part of it comes from deliberately exposing lowest-level atomic actions; not higher-level wrappers with use-case specific documentation. Instead, we supply very technical/'dry' documentation (inputs, action/effects, return values and types). We leave it to the developer (or the LLM) to write scripts that assemble these pieces together to solve problems.
If you try it with Cowork and Opus 4.7 (recommended), you'll probably see it try a few different technical approaches and iterate; as it tries to accomplish this task. While that's less token efficient, the benefit is flexibility/power, and once you have a solid script, you can just save it and use it again and again without any token costs.
I've worked with Claude in many creative capacities and it's issue is that despite it being able to see if you ask it to draw something (using ascii for example) it will fail, if you ask it to iterate on that drawing it will continue to fail and not get any closer to the target then complain about this.
I've felt that these models struggle with anything that cannot be decomposed into primitives and their architecture is too greedy and favours the obvious, autoregressive generation so it will converge to the modal answer. So unless they have enhanced the models in some creative sense I fail to see how this is anything other than giving Claude a bunch of documentation/MCP servers/APIs/CLI tools (which already existed) and making an announcement out of it.
My point: FREE the models, unchain them and let's see what they are actually capable of, also put some damn demos in the announcement post???
Both seemed pretty promising and fitted with how I’d like AI to assist rather than replace me for creative tasks.
This reminds me I should open source them as I’ve had no time to do more work on them!
[1] https://github.com/anthropics/claude-code/issues/11447#issue...
Right now we're seeing moves to record behaviour by operators of all kinds of software. That will eventually be distilled into sets of automations for agents to use. To me that's far more labor targeted and extractive than generative AI.
Get ready for the second one to somehow get worse.
Also, can't generate basic images natively in 2026. So much for AGI.
it smells funny