[_spduchamp]

If you build something and put it on the internet and have it doing it's thing by default then yes you need to be responsible for the consequences of releasing that. That's moral rights, and law needs to keep that intact.

Shoving unwanted experimental AI down everyone's throat is a bad idea and absolutely should have consequences.

Having AI peppered everywhere that can hallucinate or amplify damaging falsehoods is dangerous. AI results really should be something one specifically requests and not supersede search results as a default.

I also worry about the possible side effects of an overcorrection, like if AI companies embed a "Don't say any bad things ever" rule which would suppress critical perspectives.

Back when studying top-down AI in the 90s, everything we did was examined through a precautionary principal and liability was rule #1. It blows my mind how far we have moved away from those principals.