May 18, 2026, 3:04 PM@dzwiedziu @ai6yr The way to get a serious leash on AI is to make the AI firms fully responsible for misinformation and damage done by those systems, as humans would be who did the same thing. No 230 protection. AI likes saying "I" a lot, just like a person. They must be 100% responsible financially and criminally -- the latter meaning CEOs in prison when horrible events like suicides or murders are triggered by AI chatbots, etc.
This. ^^^
A developer is responsible for the context of a "fix" pushed under their ID. Companies are held accountable for their employees' speech, or ads they produce. The system behind it doesn't matter.
Every chat bot deployed on a customer support website gets to issue binding discounts on behalf of its controlling company.
Liability and responsibility come along with whatever benefits business owners believe they are getting from using these systems.
Firms have been held liable for their chat bots in 🇨🇦
https://www.canlii.org/en/bc/bccrt/doc/2024/2024bccrt149/2024bccrt149.html
I think that if a for-profit firm (LLM/AI slop or otherwise),
- generates, promotes or recommends something,
- claims copyright over something,
- advertises beside it to monetize it, or
- brands it,
then they are a "publisher" as specified under CDA §230(c)1 and should bear full legal liability for the content. For LLM/AI surely §230 doesn't even appear to apply.
Metcalf's Law applies: if we cripple the large commercial platforms by removing the ability to be algorithmically profitable then we remove the most powerful problems. The nasty, little players are far less powerful, so are not really the problem, IMO, and more easily policed.
@lauren
Replace "AI" with "corporations" and you'd fix a lot of flaws in capitalism. A few CEOs going to jail would do wonders to corporate governance ("pour encourager les autres"). Especially if "I didn't know" wasn't allowed as a defence (it's the CxOs' jobs to know and to insure that they know).
(This wouldn't fix all of capitalism, despite what libertarians might think; but having "skin in the game" usually improves outcomes)
@dzwiedziu @ai6yr
May 18, 2026, 4:56 PM@PeterLudemann @dzwiedziu @ai6yr That's an admirable fantasy, but that would immediately engender cries that you were trying to destroy capitalism -- it also has a decided left-wing tilt that makes it a very tough sell to the population in general. On the other hand, as I noted, hatred of LLM AI and how it's being deployed, and what it's doing to society, is absolutely nonpartisan and quite focused -- rather than just going after corporations in general. Like I said, I can't think of any other issue short of anger at child abuse that has such a nonpartisan force behind it.
@lauren @dzwiedziu Bringing responsibility to "AI" could also be a first step (a precedent) in bring responsibility to corporations. After all, the Supreme Court has said that they are "persons" in terms of speech, so why not in other aspects?
@ai6yr
May 18, 2026, 5:01 PM@PeterLudemann @dzwiedziu @ai6yr Yes, though these things need to be done in the proper order, and delicately, or you hurt the spell.
that would require repealing citizens united in the u.s.