Should such a clause not be added as standard today, similar to the “salvatory clause,” provided that the content is not intended for the widest possible distribution?
The data you produce is copyrighted by you, so it already has a very strong protection. The problem is that AI companies care shit about the law, and you’ll have to sue them in order to even get their attention. Good luck with that as a single, private person, when even bigwigs like Disney have problems here.
I do not think there is any solid answer.
The interpretation of laws is highly in flux, swinging depending on who is lobbying who on any given day.
I think we’re headed towards the worst option though, which people think they want on impulse, but actually really don’t.
I think we’re going to see regulatory capture on a scale impossible to recover from.
Regulatory capture that will simply have the biggest AI companies buy out all of the rights to all of everyones information to such a degree that it is impossible to compete with them, giving them defacto control over AI and its censorship, and everyone trying to avoid being swept up will inevitably be shovelled into using some forms of platforms that these companies have the rights to.
I think that the way it should work is that AI can input whatever it wants, and just as intended, people can claim product resultants that qualify as copywritten material.
This would both make more incentive towards having the end user run local AI/less centralization, and incentivise ownership amongst citizens.
What we’re heading to now might completely eliminate any chance we ever have of having AI that isn’t massively controlled by corporate goals and right wing billionaire feelings, and no matter how some people try to avoid it, the affect on the population will be impossible to ignore.
No clause is technically needed as you own full rights to your content by default but just like if you put something there explicitly these companies are ignoring it and the current governments are facilitating it.
The answer is legally you could and people are but these companies are just very willing to commit crimes
Yeah, the real answer is that they’d just ignore it- because even though what they’re doing is flagrant copyright abuse, unless you’ve got an army of lawyers like Disney, you’ll go bankrupt going after them.
What you can do is play dirty. If you have your own website, make an AI Tar Pit. Make your website not just useless to AI, but actively malicious towards it.
Yes, I agree. Explicitly excluding AI training in your terms of use or license probably won’t deter most companies, but I think it wouldn’t hurt to include this as an addition, since a clear contractual prohibition would likely:
- reduce ambiguities regarding defenses such as fair use and
- create an explicit basis for contractual claims that could be additionally enforced if someone ignores the restriction.
For example, when selling a book, you could require explicit consent (checkbox opt-in or similar) to strengthen enforceability. Enforcement would still be difficult, of course, but an explicit clause might at least have a certain deterrent effect and, if necessary, create additional leverage in court, I think.
Somebody already told you this is redundant. It doesn’t ‘create additional leverage’. Copyright law exists and by default grants you all these privileges. Ignorance of the law doesn’t allow someone to break it.
The problem is even if you live in a communist utopia, a lawsuit still costs time and headaches. That is the real question. Are you willing to go to court to enforce your rights? If yes, then that’s all you need to do, because you’re already the rights owner.
When the AI bots come crawling they don’t care for licensing. They will vacuum up the data and throw it into the pot of big data.
If you can reasonably prove they broke the license by sucking up your data and distributing via their models you may be able to convince a judge to award damages IF you suffered a loss.
You mean copyright? Yes, that’s good for about a hundred years and counting thanks to Disney.
Of course that only matters if you have money to take on giant corporations that have a bigger budget for their legal department than you would make in a dozen lifetimes.
That’s how it is in the US. However, I’m from Europe, where some legal systems still function reasonably well - legal disputes aren’t exactly cheap here either, but they don’t completely ruin you (can happen tho).
But yes, it’s true, you don’t stand a chance against the giant corporations here either.
Mostly no, but maybe yes. The problem would be licenses that are granted to anyone that hosts your content could allow them to use it for such things. If you completely self host or carefully select for services that don’t have terms that allow them to use your data, then maybe you have a legal claim you could make.
The second problem is going to be proving a company used your data to train on. You likely can’t prove they have a copy, if it’s been parsed into some sort of meta data, it may be permissable anyway.
We’ll know in some time - https://blog.google/technology/safety-security/serpapi-lawsuit/
You don’t need a clause if you have lawyers money. ChatGPT won’t create images with disney characters (I have tried)
We need a new html tag attribute that goes:
<tag let-ai-scrape-the-shit-out-of-this=false> content </tag>deleted by creator
