Fireship · 2026-09-15 · notable
Fireship — 'Anthropic researchers are quitting... and now we know why'
Fireship posted a video on 15 September 2026 about the wave of AI safety researchers leaving Anthropic. Jacob Coxon resigned in early September warning that labs are 'gambling with our lives', and Anthropic's own alignment lead publicly agreed.
Fireship takes on the story of Anthropic safety researchers walking out, and why the company's own alignment lead backed them up.
What is it?
'Anthropic researchers are quitting... and now we know why' went up on the Fireship channel on 15 September 2026. The video covers a run of departures from AI safety teams that began earlier in the month. Jacob Coxon, a pretraining researcher who had worked at both OpenAI and Anthropic, resigned around 9 September and said the leading labs are 'gambling with our lives' by racing toward self-improving systems.
How does it work?
The story Fireship is working from is unusual because it was not denied. TechCrunch reported that Evan Hubinger, Anthropic's own alignment lead, responded to Coxon on X by writing 'We really do earnestly believe AI could kill all humans!' and putting his personal estimate of that outcome above 10% within the next decade. That makes this an internal disagreement about pace, not an outsider's critique.
Why does it matter?
Fireship's audience is working developers rather than policy researchers, so the channel covering this pushes an AI safety argument into the feeds of people who ship code with these models every day. The departures also land while Anthropic is expected to file for an IPO, which is when the gap between what a lab says about risk and how fast it ships gets examined most closely.
Who is it for?
developers following the AI safety debate