Seems incredibly reductive and luddite. I doubt it will ever achieve adoption and projects using it will be avoided.
Not to mention that all you'd need to do is get an LLM to rewrite said programs just enough to make it impossible to prove it used the program's source code.
1. Does an AI "reading" source code that has been otherwise lawfully obtained infringe copyright? Is this even enforceable?
2. Why write a new license rather than just adding a rider to the AGPL? This is missing language the AGPL uses to cover usage (rather than just copying) of software.
Man you are thinking about using the law as your weapon. Don't want to disappoint you, but those companies/people control lawmakers. You can't fight armies of lawyers in the court.
> Any modified versions, derivative works, or software that incorporates any portion of this Software must be released under this same license (HOPL) or a compatible license that maintains equivalent or stronger human-only restrictions.
That’s not what copyleft means, that’s just a share-alike provision. A copyleft provision would require you to share the source-code, which would be beautiful, but it looks like the author misunderstood…
>without the involvement of artificial
intelligence systems, machine learning models, or autonomous agents at
any point in the chain of use.
Probably rules out any modern IDE's autocomplete.
Honestly with the wording 'chain of use', even editing the code in vim but using chatgpt for some other part of project could be argued as part of the 'chain of use'.
I think it will be interesting to see how this sort of thing evolves in various jurisdictions. I doubt it will ever fly in the US given how strongly the US economy relies on AI. US courts are likely to keep ruling that AI training is fair use because if they reversed their policy the economic consequences would likely be severe.
But EU jurisdictions? I'm quite curious where this will go. Europe is much more keen to protect natural persons rights against corporate interests in the digital sphere, particularly since it has much less to lose, since EU digital economy is much weaker.
I could imagine ECJ ruling on something like this quite positively.
In a world where AI companies cared about licenses and weren't legally permitted to simply ignore them, this might've been a good idea. But we don't live in that world.
>The idea is that any software published under this license would be forbidden to be used by AI.
If I'm reading this and the license text correctly, it assumes the AI as a principal in itself, but to the best of my knowledge, AI is not considered by any regulation as a principal, and rather only as a tool controlled by a human principal.
Is it trying to prepare for a future in which AIs are legal persons?
EDIT: Looking at it some more, I can't but feel that it's really racist. Obviously if it were phrased with an ethnic group instead of AI, it would be deemed illegally discriminating. And I'm thinking that if and when AI (or cyborgs?) are considered legal persons, we'd likely have some anti-discrimination regulation for them, which would make this license illegal.
I'm not against the idea, but licensing is a very complex subject, so this makes me think the license wouldn't hold any water against a multi billion firm who wants to use your stuff to train their AI:
> I am not a legal expert, so if you are, I would welcome your suggestions for improvements
> I'm a computer engineer based in Brussels, with a background in computer graphics, webtech and AI
Particularly when they've already established they don't care about infringing standard copyright
This is obviously not enforceable. It isn't even particularly meaningful.
Supposing the software I downloaded is scanned by a virus scanner, which is using AI to detect viruses. Who is in violation? How do you meaningfully even know when I has accessed the software, what happens if it does?
This license also violated the basic Software Freedoms. Why should a user not be allowed to use AI on software?
I've been thinking of something similar for a while now [0] except it's based on clickwrap terms of service, which makes it a contract law situation, instead of a copyright-law one.
The basic idea is that the person accessing your content to put it into a model agrees your content is a thing of value and in exchange grants you a license to anything that comes out of the model while your content is incorporated.
For example, suppose your your art is put into a model and then the model makes a major movie. You now have a license to distribute that movie, including for free...
> If you make a website using HOPL software, you are not breaking the license of the software if an AI bot scrapes it. The AI bot is in violation of your terms of service.
Assuming a standard website without a signup wall, this seems like a legally dubious assertion to me.
At what point did the AI bot accept those terms and conditions, exactly? As a non-natural person, is it even able to accept?
If you're claiming that the natural person responsible for the bot is responsible, at what point did you notify them about your terms and conditions and give them the opportunity to accept or decline?
There is too much effort going into software licensing. Copyright is not part of the meta, information wants to be free; it will always be possible to copy code and run it, and difficult to prove that a remote machine is executing any particular program.
It will get easier to decompile code as AI improves, so even the source code distribution stuff will become a moot point.
Licenses have been useful in the narrow niche of extracting software engineering labor from large corporations, mostly in the US.
The GPL has done the best job of that, as it has a whole organization dedicated to giving it teeth. Entities outside the US, and especially outside of the West, are less vulnerable to this sort of lawfare.
Ignoring the fact that if AI training is fair use, the license is irrelevant, these sorts of licenses are explicitly invalid in some jurisdictions. For example[0],
> Any contract term is void to the extent that it purports, directly or indirectly, to exclude or restrict any permitted use under any provision in
I don't know how you can post something publicly on the internet and say, this is for X, Y isn't allowed to view it. I don't think there's any kind of AI crawler that's savvy enough to know that it has to find the license before it ingests a page.
Personally, beyond reasonable copyrights, I don't think anyone has the right to dictate how information is consumed once it is available in an unrestricted way.
At a minimum anything released under HOPL would need a click-through license, and even that might be wishful thinking.
Using software is not one of the exclusive rights of Copyright holders. If I have a legitimate copy of the software I can use it, I don't need a license. Just like I don't need a license to read a book.
Open Source licenses give license to the rights held exclusively by the author/copyright-holder: making copies, making derivative works, distribution.
An open source license guarantees others who get the software are able to make copies and derivatives and distribute them under the same terms.
This license seeks to gain additional rights, the right to control who uses the software, and in exchange offers nothing else.
IANAL but I think it needs to be a contract with consideration and evidence of acceptance and all that to gain additional rights. Just printing terms in a Copyright license wont cut it.
Seriously at this point who cares about US licenses ?
It has been abuduntly clear that AI companies can train however they want, and nobody will enforce anything.
Realistically speaking, even if you could prove someone misused your software as per this license, I don't expect anything to happen. Sad but true.
At this point, I don't care about licensing my code anymore, I just want the option to block it from being accessed from the US, and force its access through a country where proper litigation is possible.
40 comments
[ 4.9 ms ] story [ 62.8 ms ] threadNot to mention that all you'd need to do is get an LLM to rewrite said programs just enough to make it impossible to prove it used the program's source code.
1. Does an AI "reading" source code that has been otherwise lawfully obtained infringe copyright? Is this even enforceable?
2. Why write a new license rather than just adding a rider to the AGPL? This is missing language the AGPL uses to cover usage (rather than just copying) of software.
> Any modified versions, derivative works, or software that incorporates any portion of this Software must be released under this same license (HOPL) or a compatible license that maintains equivalent or stronger human-only restrictions.
That’s not what copyleft means, that’s just a share-alike provision. A copyleft provision would require you to share the source-code, which would be beautiful, but it looks like the author misunderstood…
Probably rules out any modern IDE's autocomplete.
Honestly with the wording 'chain of use', even editing the code in vim but using chatgpt for some other part of project could be argued as part of the 'chain of use'.
But EU jurisdictions? I'm quite curious where this will go. Europe is much more keen to protect natural persons rights against corporate interests in the digital sphere, particularly since it has much less to lose, since EU digital economy is much weaker.
I could imagine ECJ ruling on something like this quite positively.
If I'm reading this and the license text correctly, it assumes the AI as a principal in itself, but to the best of my knowledge, AI is not considered by any regulation as a principal, and rather only as a tool controlled by a human principal.
Is it trying to prepare for a future in which AIs are legal persons?
EDIT: Looking at it some more, I can't but feel that it's really racist. Obviously if it were phrased with an ethnic group instead of AI, it would be deemed illegally discriminating. And I'm thinking that if and when AI (or cyborgs?) are considered legal persons, we'd likely have some anti-discrimination regulation for them, which would make this license illegal.
> I am not a legal expert, so if you are, I would welcome your suggestions for improvements
> I'm a computer engineer based in Brussels, with a background in computer graphics, webtech and AI
Particularly when they've already established they don't care about infringing standard copyright
Supposing the software I downloaded is scanned by a virus scanner, which is using AI to detect viruses. Who is in violation? How do you meaningfully even know when I has accessed the software, what happens if it does?
This license also violated the basic Software Freedoms. Why should a user not be allowed to use AI on software?
The basic idea is that the person accessing your content to put it into a model agrees your content is a thing of value and in exchange grants you a license to anything that comes out of the model while your content is incorporated.
For example, suppose your your art is put into a model and then the model makes a major movie. You now have a license to distribute that movie, including for free...
[0] https://news.ycombinator.com/item?id=42774179
Assuming a standard website without a signup wall, this seems like a legally dubious assertion to me.
At what point did the AI bot accept those terms and conditions, exactly? As a non-natural person, is it even able to accept?
If you're claiming that the natural person responsible for the bot is responsible, at what point did you notify them about your terms and conditions and give them the opportunity to accept or decline?
Licenses have been useful in the narrow niche of extracting software engineering labor from large corporations, mostly in the US. The GPL has done the best job of that, as it has a whole organization dedicated to giving it teeth. Entities outside the US, and especially outside of the West, are less vulnerable to this sort of lawfare.
> Any contract term is void to the extent that it purports, directly or indirectly, to exclude or restrict any permitted use under any provision in
> [...]
> Division 8 (computational data analysis)
[0] https://sso.agc.gov.sg/Act/CA2021?ProvIds=P15-#pr187-
I don't know how you can post something publicly on the internet and say, this is for X, Y isn't allowed to view it. I don't think there's any kind of AI crawler that's savvy enough to know that it has to find the license before it ingests a page.
Personally, beyond reasonable copyrights, I don't think anyone has the right to dictate how information is consumed once it is available in an unrestricted way.
At a minimum anything released under HOPL would need a click-through license, and even that might be wishful thinking.
Open Source licenses give license to the rights held exclusively by the author/copyright-holder: making copies, making derivative works, distribution.
An open source license guarantees others who get the software are able to make copies and derivatives and distribute them under the same terms.
This license seeks to gain additional rights, the right to control who uses the software, and in exchange offers nothing else.
IANAL but I think it needs to be a contract with consideration and evidence of acceptance and all that to gain additional rights. Just printing terms in a Copyright license wont cut it.
https://www.searchengineworld.com/perplexity-responds-to-clo...
It has been abuduntly clear that AI companies can train however they want, and nobody will enforce anything.
Realistically speaking, even if you could prove someone misused your software as per this license, I don't expect anything to happen. Sad but true.
At this point, I don't care about licensing my code anymore, I just want the option to block it from being accessed from the US, and force its access through a country where proper litigation is possible.
You've already violated section 1(b) by having a AI parse it, which is technically covered in fair use doctrine.
This makes it more of a philosophical statement than a functional legal instrument.