In July, the author wrote that AI is more of a threat to routine “clean” work, while messy, relationship-driven jobs like real estate brokerage stay safer because human judgment is hard to replace, citing the book Messy Jobs, The Work that AI Cannot Reach. Recently, Jacob Coxon, a researcher who spent three years on pretraining work at both OpenAI and Anthropic, resigned from Anthropic and posted a long thread on X on Sept. 8 that “neither company is acting responsibly,” maintaining there was a 10 percent chance of AI killing all of us by 2030. On Sept. 12, Anthropic CEO Dario Amodei released “We Must Pace the Frontier,” calling for a deliberate slowdown in frontier-model capability progress while safety research, testing, and governance catch up; Sam Altman of OpenAI agreed, saying he agrees with Dario that we need to pace the frontier, that this has been a primary topic of discussions at OpenAI in recent weeks, and that committing to independent evaluators with employee-like access is a great idea and OpenAI will do the same. The author is skeptical the odds of the end of civilization are this high. Jacob provided no proof, documentation or suggested solution, and made the announcement before Anthropic’s potential monster IPO, reportedly giving up an estimated $2 million. Tristan Harris is a technology ethicist known for warning about social harms of attention-driven technology and risks of rapidly deployed AI. Airplane manufacturers cannot design and release