Join the discussion
Write your take first — we'll ask for email only when you're ready to publish.
- Hacker News
- We've reached the stage where copyright has become impossible. Let's just accept it and move on.by karim79
- Heck yeah.by Cider9986
- Does this also extend to generated source code? I mean, this upend the business models of many A.I. companies like OpenAI and Anthropic. In addition, many ISVs would probably stop using A.I. if their products could lose their copyright.
And what about A.I. generated code which has been altered by a human? Is the derived work copyrightable?
by hn_submit - And this is one of the reasons some people are up in arms about Anthropic's recent inclusion of watermarking against anything Claude models touch. It will even take human-generated code that it works on and apply watermarks to it which can be then used to prove it was AI processed and perhaps invalidate any copyright on even human-written code.
- 1) Does applying a copyright message warrant that the content has not been generated by an AI?
2) If one applies a copyright message to AI generated output, is that fraudulent?
by femto - Lots of wild guesses about mixed human-AI copyright in here. Last time I read the copyright office’s statements in this, their stance was basically: The human owns exactly what they contribute. The rest is public domain.
Yes, that is vague. I think the examples were like:
If you paint a symbol and use an AI filter over that to stylize it, you own the symbol aspect of the image but not the stylized final result.
You can own a book of AI images as a curated collection. But, not the individual images.
by mindcandy - Seems about right. Copyright is supposed to literally prevent outright copying. The output from an LLM is not a creative work of the prompter. It's genuinely the opposite, I use them when I don't care about something but have to do it anyway for whatever reason. It makes more time for me to do the things I like working on.
It feels shitty that they have been trained on the life sums of all of our work and online presences with absolutely no credit given... But then again, I'm not sure I'd want to know what parts of the weights were from me and which weren't.
by dfedbeef - I don't see an easy test here. Worse, I see the beginnings of a test that is technical and very likely to not match the future of how we will interact with these things. We have to start somewhere but I'm not sure 'Neither mere prompting nor the selection between several AI suggestions is sufficient as a human creative contribution.' is the right place to start. I think we need more examples of what does constitute 'human-centric' and work from there. I also don't think that any system that says 'you didn't do enough work so it isn't human-centric' works. Any system like that will require a reexamining of copyright in general. There is a lot of low work copyrighted material out there. Just because 'AI' didn't build it doesn't mean the same tests shouldn't apply.
Some examples of why I think this is really hard: Say I build a story generation system. I work hard on building an agent swarm of actors, critics, editors, researchers. I craft into the various agents concepts of story arcs, outlining techniques, character development. I build a huge well thought out process for how to agentically write an actually good story, so long as you give it a title. Heck, I even design and train my own custom LLM with original layer ideas and novel training techniques to use on this system. After all that I then take that final step and give it a title. Do I have no claim to that? I probably put more work and creativity into it than an author would have a book. What if I then gave it 500 titles? 5,000? Would my claim degrade the more titles I fed it? Is it a percentage of work question? What is the core concept here that defines 'human-centric'? What is the cut-off here?
Let's go even further. I don't prompt. I live in a world with unlimited context models. I have a conversation about the book I want it to write. During that process I reject some ideas and accept others. I didn't give it a 'system prompt' but essentially all I did was prompt it and select versions I liked. Is that not human centric? How about if I asked it for advice and it did some editing work on my story? Did that make it not human centric even though the starting text was mine? What if that starting text was 99% replaced with a version 10x as verbose. Defining based on how you interacted with the model (prompted and selected) just seems way to weak to be a clear test.
by jmward01 - My test would be: Until the AI companies pay back what they have already stolen, then anything that uses AI at all cannot be copyrighted.by harimau777
- Information wants to be free. We should simply dispose of copyright. With LLMs, waters that were already muddy are now a thick slurry. And it's only going to get worse.
It's an antiquated mechanism and is far more abused than it is actually used at this point.
- I think there is an easy test: if you didn’t write it, it’s not yours.
If we go back 10 years and your friend says “I have an idea for an app, here it is,” and you build it, you own the copyright because you wrote it.
You give an idea to the pile of math calculated of the stolen work of humanity, the math owns it (which it can’t, so no one owns it).
No matter how detailed of a conversation you have with a friend, I don’t think they have justification to claim copyright over code written by you.
by trickypr - Does this mean someone could develop a switch emulator using an AI and Nintendo couldn't win a lawsuit against them?
Does it enable decompilation remasters of classic games?
It feels like AI is a cleanroom laundromat
- “That doesn't give you copyright, but it does give you liability for the content.”by cwillu
- Definitely not clean room, quite the opposite.by smilespray
- Im not sure it’s really a copyright issue here, and not patent or trademark related, or something else. Copyright is really only one aspect of IP laws and they all have their own legal nuancesby dgellow
- Not a lawyer, but I'd expect that decompilation creates a derived work which is still subject to the copyright of the original.by jdswain
- No, an AI's output can still violate copyright.by lg5689
- it's sort of non-controversial at the extreme end I think - where there is no human contribution, why should there be protection?
But it just begs the actual question of how much human contribution there needs to be:
- I wrote the prompt (not enough)
- I wrote many prompts and iteratively refined them using distinctly human skill (open question, but loosely seems still not enough, potentially in the EU but maybe in the US?)
- I made minor modifications post-generation (open question, probably enough)
- I made equal or more contribution to the final result (this better clearly have copyright protection or we are in real trouble)
by zmmmmm - I'd say that any use of an AI trained on people's creations without (real) consent and compensation should not be protected.by harimau777
- I think it should be at least an 80/20 rule.
80% of the code should be hand written by you. 20% can be allocated by AI for corrections or suggestions.
AI generated code should never be copyrighted otherwise. It's objectively common sense.
- > I made minor modifications post-generation (open question, probably enough)
Hmm I don't think that would be enough. I'd expect that would only make the modifications themselves copyrightable, but not the whole modified work including the AI parts.
Compare for example the case where the US copyright office ruled that, when assembling AI-generated images and human-written text into a comic book, only the human-made elements themselves (text, arrangement) get copyright protection, but not the images.
E.g. if you generate an AI photo and color grade it, i expect only the color grading would be protected (if that is even significant enough to be protectable), not the rest. And someone else could re-color-grade the same image without infringing your copyright.
by hmry - Interesting thought experiment is to consider an author who writes a novel in English, but wants to translate it into German.
They have a copyright on the original, and if they hire a human, the human would have a copyright on the translation (which would generally be licensed or transferred back to the author in some way).
If they use an AI for the translation, by the logic here, the translation wouldn't have its own independent copyright, but (based on other long established principles of copyright) it would still be a derived work of the original, so even if this decision holds it would not be legal to make unauthorised AI translations, pirate authorised AI translations, make further translations into other languages (or back to English), etc.
Which seems fairly reasonable! But consider:
If you start with, say, a 90,000 word novel, and ask for a translated novel, you (presumably) have sufficient rights to stop someone making unauthorised copies of the AI translated version.
If you start with a 300 word prompt, and ask for a logo, you (apparently) do not have sufficient rights to stop someone from using it without authorisation.
So some combination of the input (0.3k vs 90k) and the output (logo versus novel) crosses an inflection point between these two extremes, and I think it's interesting to wonder what the boundaries are. Like, in theory you could graph input size versus output complexity, and sketch a frontier between "the author's protected expression survives in the output" and "the author's protected expression does not survive in the output". And I don't have the slightest idea what I think a fair frontier would look like.
by Lazare - I don't see why the author of the 300-word prompt couldn't claim copyright of it.by matt_kantor
- Input size and output complexity are not necessarily the only two dimensions involved. There’s no reason to think that such graph would show a continuous frontier.by drdexebtjl
- > Some combination of the input (0.3k vs 90k) and the output (logo versus novel) crosses an inflection point, and it's interesting to wonder what the boundaries are
It has always been like that though. Copyright is just that arbitrary. You really can't tell if Android violates Oracle's copyright over Java by reading law text.
by raincole - This post is glossing over most of the nuance in EU law.
The AI system must function merely as a tool or instrument (like a camera or Photoshop) guided by the human, rather than acting as the creator itself. The line may get a bit fuzzy case-by-case, but effectively the human must be the creative one, not the AI.
This is not unprecedented. Machine generated technical data, sensor outputs, automated surveilance photography, monkey selfies, purely algorithmic or generative music and such were already disqualified long before AI came along.
by kstenerud - So this means you can not use AI for the majority of open source projects since GPL, MIT, BSD, etc are all copyright declarations and they'd be being made for code which you can not copyright.by socalgal2
- It will be interesting if or when someone tries to enforce that perspective. Good luck proving an online contributor used AI on a merge request.
Perhaps we'll have new iterations of FOSS licenses to adjust to legal declarations.
- wasnt there a case a while back, where GPL or LGPL code made its way into MIT licensed software via LLM? And they were forced to remove the copyleft code. I dont remember the details though.by voxelghost
- No it would mean the license does not apply to the parts of code written by AI.by randyrand