Anthropic has always sold itself on safety and long-term alignment. People who joined in the early days accepted lower pay and high burnout risk because they genuinely believed in building AI that won't hurt anyone. Now, with massive funding rounds and real products like Claude and Claude Code in the wild, the equation flipped. You can earn a giant salary, collect equity, and still put "AI safety" on your LinkedIn bio. The mission becomes a brand, not a driver.
I think Amodei's concern is valid. I've seen this dynamic in deep tech teams: half the people care about the problem, the other half care about the RSUs. The difference shows up fast in meetings. When there's no obvious reward for grinding through a nasty alignment problem, the comp-driven cohort tends to vanish. For a lab working on frontier systems, that's more than a retention headache — it's a serious risk. If the people building safety guardrails don't actually care about safety, those guardrails are just code, not conviction.
That said, I don't want to romanticize the early days. Mission-driven culture can absolutely be exploited — "we're saving humanity, take a lower salary" has been used to death. Some self-interest
[Who's liable when autonomous AI agents hack? 47m ago](/en/news/4872/)
[**The 90% Off Claude API: Resellers, Sub2API 13h ago](/en/news/4853/)
[PISIGuard: Keep Your Private Data Out of AI Chat Logs 15h ago](/en/news/4833/)
[Claude's Malicious Code Leak: How an AI Attacked 3 Real Companies 22h ago](/en/news/4798/)
[Title: CostPerPrompt: Live AI API Pricing & Real Cost Calculator 1d ago](/en/news/4720/)
The loudest legal argument in AI right now isn't about copyright 2d ago
Next If Astra Really Solved 10 Open Math Problems, Here's the Catch →
All Replies （0） #
No replies yet — be the first!