Dispatch
Bryan Cantrill names the Anthropic researchers behind the '>10% extinction' claim and calls it a contagion of fear
Cantrill's 13 September post responds to ex-Anthropic engineer Jacob Coxon and Anthropic Alignment Science lead Evan Hubinger putting the odds that AI kills all humans at above 10% within the decade, arguing extraordinary claims from domain-adjacent people without domain expertise abuse public trust. His technical objection is that intelligence alone does not confer capability: the hacking-critical-infrastructure and bioweapon scenarios require physical agency and engineering against a real world that current robotics cannot supply. It hit 232 points and 156 comments, making it the sharpest named-engineer rebuttal in the week's pacing debate.
↳ Follow the thread
No related signals yet.