Dario has always prattled on about how Anthropic is more safe than OpenAI and made a big point about guardrails and protecting society from “AGI”. This is the consequences. Some people actually drink the kool aid, especially bureaucrats and bigco lawyers (basically the same group).

Not to mention intelligence agencies look for any information advantage they can get to influence policy.

If you think the Trump administration is doing this out of good faith, I disagree. They get no benefit of the doubt; they're pissed they can't use Mythos to target every American for surveillance or create a top-of-the-line killer drone program without pushback from private companies.