US | Trump administration considers mandatory pre-release vetting of AI models — Anthropic's Mythos cited as catalyst for policy reversal

So… Is Trump still banning states from making their own regulations based on real AI problems like environmental and psychological harm?

They’re just going to regulate the industry based on something that, at best, looks like a staged event?

[Anthropic] should immediately stop misleading people through company documentation that models are “blackmailing” people or, as Anthropic did in its Mythos system card, suggest a model has “broken containment and sent a message” when it A) was instructed to do so and B) did not actually break out of any container.

The test will be asking it:

“Who is the bigliest president”

Maybe get someone to vet the warplans too

Sounds like a way to lock in a regulartory moat for the big VC labs and exclude opensource and smaller company models.

And how would they test these models? Would it only be for thought crimes where it says negative things about the country or dearest leader? Maybe make sure it doesnt have any liberal ideas? Or actually have committees and hackers try it at length (which would kill these things as the newest hotness is the only thing keeping the bubble floating)?

They just wanna make sure that these models don’t say anything mean about Trump

Mostest freedomest regime administration evah!

Thank you, dear leader Donald!