You may find https://github.com/manyfold3d/manyfold/blob/main/AGENTS.md useful ;-)
Ask HN: Is there a no-LLM license yet?
11–20 of 27 posts
Re: Ask HN: Is there a no-LLM license yet?
#12I think the problem is that even if you have a "no training" license AI companies will just ignore it and you'll be left having to take legal action. If the big publishing firms couldn't win an action against Meta for using pirated copies of their books you probably have no chance either.
Re: Ask HN: Is there a no-LLM license yet?
#13I look at this pretty similar to licenses that attempt to say something can't be used in some specific way, such as a "no evil" license. Famously, the JSON license: Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, m…
Re: Ask HN: Is there a no-LLM license yet?
#14 6. No Discrimination Against Fields of Endeavor
The license must not restrict anyone from making use of the program in a specific field of endeavor. For example, it may not restrict the program from being used in a business, or from being used for genetic research.
There's also 5. which probably isn't relevant here but possibly might depending on the definition of person (e.g. a legal person). 5. No Discrimination Against Persons or Groups
The license must not discriminate against any person or group of persons.
And then there's the obvious aspect of AI foundries blatantly ignoring copyright anyway (copyright law grants the the author the rights which he then gives away by way of a license).Re: Ask HN: Is there a no-LLM license yet?
#15If you use a non-standard license, I will pass on your project, legal is not going to take the time to review and will just reject. It can be counterproductive to your project depending on your goals You are looking for "ethical" / "permissioned" licenses (as many as there are people's gripes or causes), versus permissionless like MIT / Apache-2 / BSD-3 There is no legal way to prevent things. Bad actors will not car…
> depending on your goals is the important bit. Is your goal to have your project used by companies which have to "run things through legal", or do you have better goals?
Re: Ask HN: Is there a no-LLM license yet?
#16Some lawyers may try to weasel their way around but if it comes to a court case you should have a pretty straightforward interpretation of the phrase:
Re: Ask HN: Is there a no-LLM license yet?
#17Re: Ask HN: Is there a no-LLM license yet?
#18I think the problem is that even if you have a "no training" license AI companies will just ignore it and you'll be left having to take legal action. If the big publishing firms couldn't win an action against Meta for using pirated copies of their books you probably have no chance either.
If content with that license reaches critical mass, a law firm could sue for many at once.
Do you think book publishers are somehow less financially able to muster a legal response than open source coders in their spare time?
Re: Ask HN: Is there a no-LLM license yet?
#19I look at this pretty similar to licenses that attempt to say something can't be used in some specific way, such as a "no evil" license. Famously, the JSON license: Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, m…
How do you objectively measure if an LLM is training on something if you don't have access to its training data?
Re: Ask HN: Is there a no-LLM license yet?
#20Would this even work legally? I remember the case of books used for training, where the court found training to be fair use, but the material has to be legally obtained (=Bought instead of pirated the books). > and usage on it What do you mean by "usage on it"?
How does it work legally? IANAL so I have no idea and that's why I'm asking.