Jacob Coxon quits Anthropic as AI insiders split over what they know vs think

Alec_Coughlin · x · 2026-09-14

AIWA "T.O.T." #35 centers on Jacob Coxon's resignation from Anthropic and the debate sparked by Hilbert Spaess's exit and his beliefs about frontier AI labs. The author argues the AI news cycle has shifted again, with a core tension: frontier-lab insiders can't agree on what they actually know versus what they merely think.

The piece lays out a framework for forming your own POV: frontier researchers are literally inside the building where the electricity gets made, so their hands-on, data-driven analysis is the best starting point—but not the endpoint. Asking the right questions of the right people matters, illustrated via Clement Delangue's tweet and Benedict Evans's response to Matt Shumer's "Something Bi…" thread.

Related event: Ex-researcher's viral resignation sparks debate; Dario says he agrees more than disagrees(5 posts)→

Original post →

More from AGI Musings

AGI Musings channel →