Comment by Imustaskforhelp | Hacker News Reader

Comment by Imustaskforhelp | Hacker News Reader

Parent Full thread

Imustaskforhelp

This is really interesting. I have some questions though, (as I said in every comment here, fair disclaimer: IANAL)

if someone used your art put it into a model and makes the major movie, you now have a license to distribute that movie, including for free...

What about the Model itself though, it is nothing but the weights which are generated via basically transforming the data that was unlawfully obtained or one which actually violated the contract-law

it wasn't the person creating the prompt which generated the movie via the model , it wasn't either the movie or the prompt which violated the contract but the model or the scraping company itself no?

Also you mention any output, that just means that if someone violates your terms of service and lets say that you created a square (for lack of better words) and someone else created a circle

and an ai is trained on both, so it created both square and circle as its output one day

What you say is that then it should give you the right to "use and re-license any output or derivative works created from that trained Generative AI System."

So could I use both square and circle now? Or could I re-license both now? How would this work?

or are you saying just things directly trained or square-alike output would be considered in that sense

So how about a squircle, what happens if the model output a squircle, who owns and can relicences it then?

What if square party wants to re-license it to X but circle party wants to re-license it to Y

Also what about if the AI Company says its free use/derivative work, I am not familiar with contract law or any law for that matter but still, I feel like these things rely an underlying faith in the notion that AI and its training isn't fair work but what are your thoughts/ how does contract law prevent the fair work argument

> data that was unlawfully obtained or one which actually violated the contract-law

This is indeed a weak-point in the contract approach: People can't be bound by an contract they never knew about nor agreed-to.

However if they acquired a "stolen" copy of my content, then (IANAL) it might offer some new options over in the copyright-law realm: Is it still "fair use" when my content was acquired without permission? If a hacker stole my manuscript-file for a future book, is it "fair use" for an AI company to train on it?

> it wasn't the person creating the prompt which generated the movie via the model

The contract doesn't limit what the model outputs, so it doesn't matter who to blame for making/using prompts.

However the model-maker still traded with me, taking my stuff and giving me a copyright sub-license for what comes out. The "violation" would be if they said: "Hey, you can't use my output like that."

> So could I use both square and circle now? [...] a squircle

Under contract law, it doesn't matter: We're simply agreeing to exchange things of value, which don't need to be similar.

Imagine a contract where I trade you 2 eggs and you promise me 1 slice of cake. It doesn't matter if you used those eggs in that cake, or in a different cake, or you re-sold the eggs, or dropped the eggs on the floor by accident. You still owe me a slice of cake. Ditto for if I traded you cash, or shiny rocks.

The main reason to emphasize that "my content is embedded in the model" has to do with fairness: A judge can void a contract if it is too crazy ("unconscionable"). Incorporating my content into the their model is an admission that it is valuable, and keeping it there indefinitely justifies my request for an indefinite license.

> What if square party wants to re-license it to X but circle party wants to re-license it to Y

If the model-runner generates X and wants to give square-prompter an exclusive license to the output, then that's a violation of their contract with me, and it might be grounds to force them to expensively re-train their entire model with my content removed.

A non-exclusive license is fine though.

To state the obvious, IANAL.

> This is indeed a weak-point in the contract approach: People can't be bound by an contract they never knew about nor agreed-to.

"Prominent notice" is important in the terms of use approach. Many terms of use claims have been dismissed because there was failure to give prominent notice - however, there have been successes, such as Hubbert vs Dell[1], where an appeals court reversed the trial court's decision and ruled in Dell's favour on the basis that they had given prominent notice of terms.

[1]:https://caselaw.findlaw.com/court/il-court-of-appeals/124479...

There are other potential legal avenues besides contract law, such as Unjust Enrichment[2], which, according to Wiki is analysed as:

    1. Was the defendant enriched?
    2. Was the enrichment at the expense of the claimant?
    3. Was the enrichment unjust?
    4. Does the defendant have a defense?
    5. What remedies are available to the claimant?

[2]:https://en.wikipedia.org/wiki/Restitution_and_unjust_enrichm...

Since AI companies are likely to be enriched (by a large amount), and it could be at the expense of a claimant if the AI (re)produces the claimants work or closely related work based on it (or in the case of a class action suit, potentially many works), the AI companies which violate a terms of use would have to argue in their favour for #3 and #4 - is the enrichment unjust and do they have a defense.

There are certainly arguments for AI trained on copyrighted works being unjust. Including a terms of use which specifically prevents this would be in the favor of the claimant. An AI company would have to defend their decision to ignore such terms and claim that doing so is not unjust.

Arguably, if the AI is sufficiently "intelligent" enough, it should be able to make sense of such terms and be used to filter such works from training data (unless specifically prompted to ignore). If the AI companies are not filtering the data they aggregate then there's a potential argument for negligence.

There's some efforts being made, such as the IETF aipref[3], which could standardize the way training data is collected/filtered by AI companies. Creative Commons have a related effort called "CC signals". These could also be helpful in a future claim if the AI companies ignore those preferences.

[3]:https://datatracker.ietf.org/wg/aipref/about/

So to me it seems having a clause in your license and/or terms of use which prevents use in AI training, providing prominent notice of such terms, and indicating AI preferences in the protocols, is going to be better than not having such clauses - because if you don't have the clause then the defendant can simply claim in their defense that "There was nothing in the terms which said we can't".

It's up to us to take a proactive approach to protecting our works from AI, even if there is not yet any solid legal basis or case law, so I applaud OP's efforts even if the license doesn't have carry any weight. If we don't attempt to protect our works from use AI training then the AI companies will win at everyone else's expense.

Creators should decide how their works are used, and we need to normalize it being the case that they decide whether or not use in AI training data should be permitted.

Reply on news.ycombinator.com