I resigned from Anthropic today (Jacob Coxon)
20 days ago
- The author resigned from Anthropic after three years in pretraining research at OpenAI and Anthropic, citing both companies' irresponsible race toward self-improving superintelligence.
- The technology is advancing rapidly toward superhuman systems capable of hacking, revolutionizing fields, and acquiring real power, with no signs of slowing down.
- AI builders genuinely believe the technology could kill everyone by the end of the decade, expressing fear privately despite downplaying it publicly.
- At OpenAI, many haven't internalized the stakes; at Anthropic, they understand the risks but are locked in a race, believing only they can act responsibly.
- The 'endgame' approach of racing ahead is a hubristic gamble that should not be initiated by a private company without extraordinary confidence in alignment.