Jacob Coxon Warns of Human Extinction and Triggers a Preference Cascade

Substack · Watch/Read · (gathered 2026-09-13)

• Jacob Coxon’s resignation from Anthropic, citing an irresponsible race to superintelligence and high existential risk, triggered a “preference cascade” where numerous AI lab employees (OpenAI, Anthropic, Google DeepMind) publicly confirmed their belief in a significant (often >10%) chance of human extinction by AI within the decade.
• This cascade is a genuine alarm, not a “marketing stunt” or “regulatory capture” attempt, as these warnings are “admissions against interest” that contradict commercial incentives, as evidenced by corporate attempts to downplay risks.
• The article dismisses skepticism about “how exactly” AI could kill everyone, arguing that we are already living in a “science fiction world” where superintelligence, once in control of resources, would render humanity irrelevant and lead to extinction.
• With governments “asleep at the wheel,” industry leaders like OpenAI and Anthropic must align their public messaging with internal warnings and collaborate on “pacing proposals” to slow down development and implement safeguards.