The contagion of fear
Points and comments are a snapshot, not live.
An engineer argues that AI extinction claims from ex-Anthropic staff are irresponsible fear-mongering.
Bryan Cantrill recounts a shameful prank from his university years where he falsely claimed a computer virus was spreading, causing panic and lost work. He draws a parallel to recent claims by ex-Anthropic employee Jacob Coxon and Anthropic's Evan Hubinger that AI has a >10% chance of killing all humans within a decade. Cantrill rebuts that engineering requires physical action and human accountability, not just intelligence. He argues domain experts should not abuse public trust by making extraordinary claims without commensurate evidence, especially regarding extinction.
What commenters are saying
Commenters split into two camps. Supporters of the article praised it as a responsible call for evidence, noting that fear leads to irrational reactions and that the precautionary principle is rarely applied in practice. Critics argued the article dismisses real risks: they cited actual AI incidents (agents hacking infrastructure, collective behavior) and claimed the author's rebuttal (robots aren't capable enough yet) ignores remote threats and recursive self-improvement. One commenter compared AI safety to pharma CEOs admitting unknown catastrophic risks, highlighting the precautionary principle.