What happened
Jacob Coxon resigned from Anthropic this week, warning the company and OpenAI were "gambling with our lives" in a race toward superintelligence. His post exceeded 100 million views within days.
Why it matters
Jan Leike's near-identical OpenAI resignation post drew 6.1 million views in May 2024, and Mrinank Sharma's Anthropic exit drew roughly 1 million in February 2026. Coxon's post landed after data-center opposition became a majority.
What to watch
Fortune's reading is that the reach reflects timing, not a stronger argument. A recent Reuters/Ipsos survey found 64% saying it is not "a good thing to build data centers at a rapid rate," with 77% worried about electricity rate increases.
WHO IT HITSThis matters most to people tracking AI policy and data-center siting — governors, local planners, and the tech leaders who privately fear a "lasting political crisis," according to Politico. It also lands on the AI-safety researchers at Anthropic and OpenAI who publicly backed Coxon's account.
Summaries like this, every morning.
Fortune frames the Coxon episode not as a new argument but as the fifth notable safety-motivated exit or mass statement from frontier labs since 2023, following Geoffrey Hinton's departure from Google in May 2023, Ilya Sutskever's and Jan Leike's resignations from OpenAI in May 2024, Mrinank Sharma's exit from Anthropic in February 2026, and a July 2026 open letter signed by more than 1,100 employees.
The difference, per the article, is the political ground it landed on. Independent tracking by Heatmap, Echelon Insights, and Gallup identified February and March 2026 as the months public opinion on data centers turned, moving into majority opposition for the first time. By September, a Reuters/Ipsos survey found 64% against rapid data-center buildout.
Fortune also offers a less flattering explanation: the availability cascade described by economists Timur Kuran and Cass Sunstein, in which a claim seems more plausible simply because it keeps appearing. The article notes this framework does not require the claim to be false, so the size of Coxon's reach is not by itself evidence that his warning is more credible than Sharma's or Hinton's.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Fremantle and RoseBerry Media will use AI to scan decades of Neighbours footage and make short vertical micro-…
Sources say an Accenture employee gave their personal Claude Code account administrative access to a DOD NIPRN…
Mistral released the weights for its trillion-parameter Mistral Large 4 ("Le Chonk") on Hugging Face and via i…
OpenAI updated its Model Spec around five values — faithfulness to human intent, harm minimization, honesty an…
OpenAI replayed the Hugging Face incident with its expanded misalignment monitors, which flagged the agents' s…
Meta deployed Catching JiTTests — LLM-generated tests designed to fail on buggy code — and says they caught 74…
