Anthropic's Jacob Coxon quits over AI 'gambling with our lives'

The resignation of Anthropic researcher Jacob Coxon is not an isolated event. Within Anthropic itself, the alignment science lead agreed with the possibility that AI has a more than 10% chance to 'kill all humans' in the next decade. OpenAI's chief scientist also argued for a research slowdown, and CEO Sam Altman reportedly told staff the company is considering slowing leading-edge AI work, provided the government says that's not an antitrust problem. The article frames these as pieces of a long-smoldering argument rather than a coordinated shift.

The disconnect is the story's real thread. In the same week, Meta, OpenAI, DeepSeek and others released a flurry of new models, agents and apps. The article's author says they are still somewhat skeptical of the doomers, given the lack of specific reasons why AI cannot be controlled, but acknowledges the contradiction. A separate force that may slow things down regardless is rising opposition to data centers, which the article says leaves tech giants with a lot of work persuading communities.

Where this leaves the reader is uncertain. The article does not say whether the safety talk will translate into action, and it notes that we will have to see if any of this existential worry changes anything. For AI lab employees and leadership, the question is whether internal concern translates into changes in release pace. For communities weighing data center projects, the near-term friction may matter more than the philosophical debate.