If you use AI-generated code, you currently cannot claim copyright on it in the US.
-
@jamie @starr Registration is required to sue to enforce a copyright, yes. The copyright exists without registration, but as soon as you want to sue, you have to provide the registration number or a copy of the Register of Copyright's formal denial (which you can then litigate).
What registration gives you is "statutory damages": for infringement that occurs prior to registration, you can only receive *actual* damages, not the act's fixed penalty plus treble damages and costs.
@jamie @starr This was a big deal for authors in the Anthropic suit: those whose works had not been registered for whatever reason prior to the infringement were excluded from the settlement because they would only have been entitled to at most a few dollars in lost royalties, a fact-bound question not conducive to class action and for which they could not be awarded fees. (Foreign authors are understandably angry about this.)
-
@starr Also, are the full contents of all registered copyrights visible at the Library of Congress? I assumed that was patents only but I used to get copyright and patents confused a lot and this may be one of those things I've been carrying incorrectly in my mind.
@jamie @starr No. You must deposit the work with the Copyright Office but the rules vary depending on the kind of work and the nature of the claim. For very voluminous non-literary works, the Office has long allowed deposit of a representative sample. While the Copyright Office is part of the Library of Congress, copyright deposits do not become part of the Library's public collections. (The Librarian can require publishers to deposit copies of specific works for public access.)
-
@jamie It may not be copyrightable but it can still count as one of the other forms of IP, such as trade secrets, which as a former employee you can very easily be prosecuted for divulging. And there are usually NDAs on top of that. (sorry, replied to wrong post before)
@dwineman Yeah, there are a few approaches to IP law that they'd still have available. But even then, the discovery process would probably require them to air out some of their laundry, too.
IIRC one of the AI platforms was sued recently and settled out of court. Someone on here pointed out that they likely did that to avoid discovery, which would enter a lot of internal data into public record. I'm fuzzy on the details, but the gist was companies generally don't like to go to court over this.
-
@dwineman Yeah, there are a few approaches to IP law that they'd still have available. But even then, the discovery process would probably require them to air out some of their laundry, too.
IIRC one of the AI platforms was sued recently and settled out of court. Someone on here pointed out that they likely did that to avoid discovery, which would enter a lot of internal data into public record. I'm fuzzy on the details, but the gist was companies generally don't like to go to court over this.
@jamie The problem is that as an individual, that process would likely bankrupt you well before it even got to discovery, and the company knows that.
-
@jamie The funny thing about this whole thread is apparently I'd already blocked that guy some time ago, so I'm only seeing your side of the conversation. And…that's all I need to know anyway.
@jaredwhite Yeah, you didn't miss much. Mainly he was replying to things I wasn't saying. Easiest argument I've had on the internet in years.
-
@jamie The problem is that as an individual, that process would likely bankrupt you well before it even got to discovery, and the company knows that.
@dwineman 100%. They don't need a favorable judgement to silence you.
-
If you use AI-generated code, you currently cannot claim copyright on it in the US. If you fail to disclose/disclaim exactly which parts were not written by a human, you forfeit your copyright claim on *the entire codebase*.
This means copyright notices and even licenses folks are putting on their vibe-coded GitHub repos are unenforceable. The AI-generated code, and possibly the whole project, becomes public domain.
Source: https://www.congress.gov/crs_external_products/LSB/PDF/LSB10922/LSB10922.8.pdf
@jamie RIP microsoft
-
Stop whining. You and about seventy zillion terrified sheep running around here bleating about the Terrible AI monster under the bed.
@tuban_muzuru @jamie as a random viewer of this thread, you come off as utterly insufferable, which might not be what you think you come off as, and so you might want to reconsider your behavior
-
If you use AI-generated code, you currently cannot claim copyright on it in the US. If you fail to disclose/disclaim exactly which parts were not written by a human, you forfeit your copyright claim on *the entire codebase*.
This means copyright notices and even licenses folks are putting on their vibe-coded GitHub repos are unenforceable. The AI-generated code, and possibly the whole project, becomes public domain.
Source: https://www.congress.gov/crs_external_products/LSB/PDF/LSB10922/LSB10922.8.pdf
@jamie it's the same in Germany. You can't copyright anything that isn't created by a human.
-
If you use AI-generated code, you currently cannot claim copyright on it in the US. If you fail to disclose/disclaim exactly which parts were not written by a human, you forfeit your copyright claim on *the entire codebase*.
This means copyright notices and even licenses folks are putting on their vibe-coded GitHub repos are unenforceable. The AI-generated code, and possibly the whole project, becomes public domain.
Source: https://www.congress.gov/crs_external_products/LSB/PDF/LSB10922/LSB10922.8.pdf
@jamie Anything AI-generated is free, BUT anything AI-generated is also worse than simply worthless.
*shrug* -
If you use AI-generated code, you currently cannot claim copyright on it in the US. If you fail to disclose/disclaim exactly which parts were not written by a human, you forfeit your copyright claim on *the entire codebase*.
This means copyright notices and even licenses folks are putting on their vibe-coded GitHub repos are unenforceable. The AI-generated code, and possibly the whole project, becomes public domain.
Source: https://www.congress.gov/crs_external_products/LSB/PDF/LSB10922/LSB10922.8.pdf
@jamie, so if a code review agent corrects a variable name in a proprietary 5M LOC project and that AI edit is not documented (where?), the entire project becomes public domain?
(Asking not you specifically but to entertain the thought such a law could be written without nuance.)
-
If you use AI-generated code, you currently cannot claim copyright on it in the US. If you fail to disclose/disclaim exactly which parts were not written by a human, you forfeit your copyright claim on *the entire codebase*.
This means copyright notices and even licenses folks are putting on their vibe-coded GitHub repos are unenforceable. The AI-generated code, and possibly the whole project, becomes public domain.
Source: https://www.congress.gov/crs_external_products/LSB/PDF/LSB10922/LSB10922.8.pdf
@jamie Hmm... this sounds like it's saying is that if a work (e.g. a code base) includes AI-generated content and doesn't identify which parts of it were AI-generated, the whole work and every part of the work, even the human-created parts, become ineligible for copyright. I believe that's wrong. (Maybe misleading, at best, if you meant it in some other way.) I mean, I can't say so authoritatively, since I'm a copyright nerd, not a lawyer, but I'm becoming increasingly convinced the more I look into it. If nothing else, it'd be a quick way to invalidate anybody's copyright on anything by just combining it with some AI-generated content and releasing the combination.
I think a more accurate statement would be that if you fail to disclose which parts were not written by a human, the copyright status of the work is unclear. The human contributions are still copyrighted by their authors, but there are some things that can't be done with the work as a whole without knowing which contributions those are.
-
@fsinn @jamie My understanding was that training an AI model on copyrighted work was fair use, because the actual "distribution"--when the AI generates something from a prompt--uses a diminimus amount of copyrighted content from an individual work, except if the user explicitly prompted something like, "Give me Homer Simpson surfing a space orca," at which point the AI company would throw the user all the way under the bus.
-
If you use AI-generated code, you currently cannot claim copyright on it in the US. If you fail to disclose/disclaim exactly which parts were not written by a human, you forfeit your copyright claim on *the entire codebase*.
This means copyright notices and even licenses folks are putting on their vibe-coded GitHub repos are unenforceable. The AI-generated code, and possibly the whole project, becomes public domain.
Source: https://www.congress.gov/crs_external_products/LSB/PDF/LSB10922/LSB10922.8.pdf
@jamie@zomglol.wtf Is Windows FOSS now?
-
If you use AI-generated code, you currently cannot claim copyright on it in the US. If you fail to disclose/disclaim exactly which parts were not written by a human, you forfeit your copyright claim on *the entire codebase*.
This means copyright notices and even licenses folks are putting on their vibe-coded GitHub repos are unenforceable. The AI-generated code, and possibly the whole project, becomes public domain.
Source: https://www.congress.gov/crs_external_products/LSB/PDF/LSB10922/LSB10922.8.pdf
@jamie so windows 11 source code is public domain now?
What about AWS? -
@Azuaron @fsinn The argument has been that the model doesn't contain the copyrighted works directly. Like, you can't grep the model file on disk for a passage from a book it can still somehow reproduce.
It's a ridiculous argument, though, because the models deal in numbers, not text. Those numbers are converted to text for human consumption only, so of course it won't contain the raw text anywhere in the model.
-
@Azuaron @fsinn The argument has been that the model doesn't contain the copyrighted works directly. Like, you can't grep the model file on disk for a passage from a book it can still somehow reproduce.
It's a ridiculous argument, though, because the models deal in numbers, not text. Those numbers are converted to text for human consumption only, so of course it won't contain the raw text anywhere in the model.
-
@max @fsinn @jamie That's not true. Media organisations and individual journalist make a share of their income from granting licenses for secondary use of their digital works, for copying them or for offering them in libraries. Copyright is one of the few bedrocks of income. It doesn‘t vanish through wishful thinking or ignoring it.
-
@jamie I *am* an IP lawyer and I (along with many others) have been saying it for a while, that if the position the “AI” co’s are taking with respect to the legality of scraping “publicly available” materials were true (that all “publicly available” materials are “public domain” free to be used as raw materials without consent required), then copyright ceases to exist and all their own materials will be free for everyone else to use the very first time they’re leaked. That’ll be fun for the co.
@fsinn @jamie Thanks! Obtaining copyright for LLM-generated text is one thing, but I've read an assessment from a German state ministry yesterday that according to national laws copyright infringement by LLMs are passed on to users and text they generate in Germany, in their interpretation. If that holds, consequences might be rather big.
-
FWIW I'm not a lawyer and I'm not recommending that you do this.
Even if companies have no legal standing on copyright, their legal team will try it. It *will* cost you money.But man, oh man, I'm gonna have popcorn ready for when someone inevitably pulls this move.
@jamie Hopefully they won't. If you right now published your company's non-AI code, you can be sure copyright infringement won't the thing that kills you, that's just a cherry on top.
So if you do it with a codebase that has undisclosed AI code, you're still ruining your life except they won't have their cherry on top. Not sure it's worth it but YMMV.
