There is a lot of legal history of doing new designs with influence from reverse engineered sources. The basic rule is that you can't have someone familiar with the original code involved with writing the new code.
LLMs are a new area, but I would expect that feeding the original source into an LLM context and saying "Write this, but different" wouldn't qualify as a clean-room design. There isn't enough general removal from the original subject matter.
Specifically about "humans": my understanding is that Clean-room design is not a requirement and the "Case law" section on your Wikipedia link explains that and has examples.
How and whether the same principles can be applied to LLMs, I have no idea. I imagine it would involve discussions about creativity, for example.
The thing that happened with "clean-room design" is that one company did it and the court said it was allowed. Companies are risk averse. Once they know one thing is yes and something else is maybe, they go with the yes, right? With the result that the supply of court cases where something other than that happened tends to dry up, leaving the other alternatives to stay as maybe.
If some people have now decided they're not going to blink in the face of a risk, we might get to find out whether some of the maybes turn out to be a yes or a no.
I have done this. Despite using Fable for the spec, and having it double-check the spec a couple of times, it still wasn't airtight. And at some point, the implementing LLM accessed one file of the original code on the Web.
I have by now deviated quite a bit from the original (well, original...) spec.
Would have to be proven in court. Look up history of cases and settlements for GPL violations and it quickly becomes apparent that it’s relatively toothless and mostly license theater.
In my opinion yes. Just like if you watch a Bob Ross episode and copy all his brushstrokes, you have made a derivative (in the literal sense of the word) of his painting.
Not content to destroy one potential competitor, USL also seeks to keep a stranglehold on the entire market by claiming that everyone who has had ever had access to UNIX source code a group that include~ most experienced programmers in the industry is "mentally contaminated" and may not develop any functionally similar product, without first purchasing an exorbitantly expensive license from USL. In accordance with this argument, USL requests this Court to prohibit BSDI from hiring anyone who has had access to UNIX source code. USL's request would not only require BSDI to fire all of its current employees, but would make it difficult to find any qualified replacements. Having chosen to widely disseminate its source code in order to build a market for its product and having succeeded in making it an industry standard, USL cannot now claim that everyone with knowledge of it is restricted from developing competing products.
Mechanical transformation is always considered derived.
The difference is that if humans do it, there's plausible deniability that it is derived.
I mean, even when humans do it, we still do clean room engineering.
It goes further though: LLMs are just a tool, and tools don't get rights, humans do.
Humans having the right to learn from a few dozen books and projects is not the same as a tool that infers output using all copyrighted works know to man.
Yeah, absolutely. The defense there is "clean room reimplementation ". If anything, the use of an LLM would be stronger than a human doing the copying. (Assuming American jurisdiction!)
https://en.wikipedia.org/wiki/Clean-room_design
There is a lot of legal history of doing new designs with influence from reverse engineered sources. The basic rule is that you can't have someone familiar with the original code involved with writing the new code.
LLMs are a new area, but I would expect that feeding the original source into an LLM context and saying "Write this, but different" wouldn't qualify as a clean-room design. There isn't enough general removal from the original subject matter.
Specifically about "humans": my understanding is that Clean-room design is not a requirement and the "Case law" section on your Wikipedia link explains that and has examples.
How and whether the same principles can be applied to LLMs, I have no idea. I imagine it would involve discussions about creativity, for example.
Not a lawyer.
The thing that happened with "clean-room design" is that one company did it and the court said it was allowed. Companies are risk averse. Once they know one thing is yes and something else is maybe, they go with the yes, right? With the result that the supply of court cases where something other than that happened tends to dry up, leaving the other alternatives to stay as maybe.
If some people have now decided they're not going to blink in the face of a risk, we might get to find out whether some of the maybes turn out to be a yes or a no.
would using one LLM to create a spec, and another with only access to the spec doc count as clean enough?
Only if the second LLM can be proven to never got trained/distilled from a model that trained on the original source code.
Probably impossible if the original code is publicly available since most code oriented LLM just gobble them all.
I have done this. Despite using Fable for the spec, and having it double-check the spec a couple of times, it still wasn't airtight. And at some point, the implementing LLM accessed one file of the original code on the Web.
I have by now deviated quite a bit from the original (well, original...) spec.
Would have to be proven in court. Look up history of cases and settlements for GPL violations and it quickly becomes apparent that it’s relatively toothless and mostly license theater.
In my opinion yes. Just like if you watch a Bob Ross episode and copy all his brushstrokes, you have made a derivative (in the literal sense of the word) of his painting.
https://www.paulaoki.com/.admin/930108.oppose.html
https://en.wikipedia.org/wiki/UNIX_System_Laboratories,_Inc.....Mechanical transformation is always considered derived.
The difference is that if humans do it, there's plausible deniability that it is derived.
I mean, even when humans do it, we still do clean room engineering.
It goes further though: LLMs are just a tool, and tools don't get rights, humans do.
Humans having the right to learn from a few dozen books and projects is not the same as a tool that infers output using all copyrighted works know to man.
It could be the same act - but not necessarily legally the same because of the actor.
Yeah, absolutely. The defense there is "clean room reimplementation ". If anything, the use of an LLM would be stronger than a human doing the copying. (Assuming American jurisdiction!)
Clean room implementation is not a concept in copyright law. It's a defensive legal strategy to avoid expensive lawsuits.
Copyright law does not require a clean room implementation when copying non-copyrightable elements of a work, such as algorithms.