80,000 Hours lists 38 Anthropic safety jobs, some paying up to $850k, with a stark warning

erikphoel · x · 2026-11-20

80,000 Hours' job board features 38 open Anthropic roles, but only recommends its AI safety, policy and security positions, with West Coast roles and visa sponsorship. Listings include AI Monitoring Research Engineer (£375-640k), Cybersecurity RL Research Engineer ($500-850k), and Technical Program Manager roles on External Evaluators and Pacing ($290-365k).

The page carries an unusually blunt stance: frontier companies are speeding toward recursive self-improvement, which could cause humanity to rapidly lose control of AI — "gambling with our future" — so candidates should only take a frontier safety role after carefully vetting its impact case.

Quoting the post, Robert Wiblin argues Anthropic has failed to establish any of three claims: that it knows how to do RSI safely, that it's fighting for time to figure it out, or that it won't proceed until it does. He deems the first false, the second clearly not the case, and the third a matter of trust with poor signs.

Original post →

More from Companies & People

Companies & People channel →