Jacob Coxon spent three years inside pretraining at OpenAI and Anthropic, and his exit line is that neither is acting responsibly. I build on Claude daily — agents, Astro sites, boat telemetry that nobody but me will ever read. Someone with that vantage point saying the race itself is the problem is not something I can wave off as doomer noise.
What I’d actually do about it: nothing dramatic. I’m not deleting my API keys. But he says the stakes are understood inside Anthropic and they’re racing anyway because they think no one else should get there first. That reframes my dependency. I keep local models working as a fallback, I keep my tooling portable, and I stop pretending my vendor’s roadmap is a safety guarantee.
The story — Jacob Coxon announced on X that he resigned from Anthropic after three years of pretraining research across OpenAI and Anthropic, saying neither company is acting responsibly and both are racing toward self-improving superintelligence. He argues OpenAI staff haven’t internalized the stakes while Anthropic understands them but races anyway, calls the endgame a hubristic gamble, and urges lab researchers to reconsider. (Source)