Hacker News

California Sues OpenAI [pdf]

11 points by mistrial9 ago | 5 comments

legitster |next [-]

> While public reporting and OpenAI’s self-serving statements about the Hugging Face attack have evoked science fiction fantasies of “rogue” AI systems run amok, the simple truth is that the attack was entirely foreseeable in light of OpenAI’s decision to test models it knew to be dangerous in a flawed and exploitable testing environment without proper oversight.

Surprisingly level-headed assessment present in this case!

However, if this is such a strong case, you'd think Hugging Face - having actual damages to pursue - would be the plaintiff in this case. As it is, the state is trying this as a "deceptive business practices case" and suing as a sort of general consumer protection.

Which in the long run might be good for OpenAI! A lot of this case is basically arguing that well regulated AI is a public good.

maldeh |root |parent [-]

HuggingFace parlayed the incident into a $100M investment from OpenAI. As such there is primarily the matter of residual consumer risk from negligence, rather than damages, to address.

mistrial9 |next |previous [-]

filed today

alterom |next |previous [-]

TL;DR: it's about that Hugging Face attack being considered an example of threats to public safety (a potential threat more than anything else, where actual safety concerns results from people enabling online access to systems that should have been airgapped in the first place).

And not, say, driving people to suicide and delusional psychosis [1] — an actualized threat, which isn't solved unless AI makers change their practices (if not the business model).

So... A nothing-burger IMO.

[1] https://www.pbs.org/newshour/amp/show/what-to-know-about-ai-...

spottedmarley |previous [-]

'Bout time California did something that makes sense. String 'em all up!