This should be quite interesting to watch. It seems widely agreed that the "hands off" approach worked well with the internet. Only now after decades are some people trying to turn the screws with section 230 repeals, age verification, and all sorts of other onerous requirements on internet usage that were never needed before.

I wonder how badly these onerous requirements on AI will stunt its growth.

Bannon is not good at knowing which fights to pick. Everyone is on the same side. He could be saying, work for us exclusively or we will make with the regulations; but he's not that smart, and that train's not even leaving the roundhouse. AI is now all and only about the forthcoming bailouts.

This must be some kind of clever trick by the Europeans and Chinese to hobble AI in the US.

More regulation, more red tape and less economic growth.

And then in a years time the Chinese models will have these capabilities anyway.

Realistically, how much "vetting" can they do? If a few people trying to poison the "language" in LLMs doesn't matter, surely even Trump can't get an LLM to be more conservative.

There’s part of me that doesn’t hate this idea, but If there’s anything I’ve learned about Trump it’s that his definition of conservative is 180 degrees from mine.

It will be less classical liberalism / burkian conservatism and just more anti-immigrant hate. Less worker protections, more power consolidation.