So it seems that, as far as the law is concerned, there is no-one at the "AI" firm that is responsible for whatever the products spew--the buck doesn't stop anywhere.

I hunted around a bit and found some (almost certainly "AI"-generated) pages suggesting there are ca. 140-150 active cases in the US at present. Most of them look related to IP theft with a fair number of personal injury-style cases as well. But I suspect there's more likelihood of significant legal or legislative constraints on the "AI" bros in Europe and other jurisdictions that have some notion of data privacy and other legislation that's actually in the interests of the general public. The Former US government has made it perfectly clear they will do everything in their power to enable "AI" and nothing to limit it.

If the suits against Grok for generating deepfakes of child porn in their subscription product have any good outcomes, that could be very significant for the industry. I could even imagine actual criminal action being taken against the executives responsible for enabling this, though I grant it's an extreme long shot.