trump’s new best friends keep crashing the simulation
The White House is particularly charmed by Anthropic, whose CEO Dario Amodei hung out with Trump in late September while somehow forgetting to mention that his models had just caused "safety incidents" on government websites. The company waited until October 8 to tell officials, then dropped a Friday-night blog post like a teenager hiding a bad report card. They followed that up with an AI "safety" bill that both embraced federal preemption of state AI laws and decided catastrophe starts at 50 deaths or more, then walked it back when everyone noticed they’d said the quiet part out loud. Regulators now trust them about as much as Trump trusts a prenup.
OpenAI isn’t exactly covering itself in glory either. Its chief lobbyist is out here posting vague policy proposals that magically trend toward blocking states from regulating, while a growing public majority says AI is bad for society and the FTC circles both companies like a very bored shark. Meanwhile, Democrats are licking their chops at Trump snuggling up to tech billionaires as AI threatens jobs, workers, and kids—because nothing screams "man of the people" like doing photo-ops with guys who think a "manageable" body count is just part of innovation. The emerging bipartisan consensus for AI CEOs is simple: talk less, disclose more, and maybe stop treating democracy as a beta test.











