Anthropic likes to think of itself as the "good guy" of AI — the lab that would never let an OpenAI-style Sam Altman go off the rails. But according to former employees, that carefully curated image is cracking. The company, which proudly brands itself around "high trust, low ego," is sliding into intellectual groupthink.
Multiple ex-staffers tell WIRED that harsh internal criticism is mostly confined to private group chats. Regularly scheduled all-hands meetings have earned the ironic nickname "Dario Vision Quests," after CEO Dario Amodei. Listening to Amodei talk, one former employee says, is like "hearing a sermon." Openly challenging management at company-wide meetings? Rarely happens.
That subdued atmosphere, insiders say, comes from an elite security mindset Anthropic has embraced: "offense as defense."