A senior safety researcher at Anthropic has stated there is more than a 10 percent probability that artificial intelligence could eradicate humanity before the decade ends. This warning emerged just hours after a fellow employee resigned, citing alarming safety lapses within the company and its competitors.
Jacob Coxon, who previously worked on AI systems at OpenAI, announced his departure from Anthropic in a post on X. He accused both firms of recklessly racing toward the development of self-improving superintelligence, describing the move as a dangerous gamble with human survival.
So the smartest people in the room think we might be doomed? Thanks for the weekend peace of mind, guys.
The industry is definitely racing ahead without enough safety brakes. This resignation highlights a serious cultural problem inside these labs.
How can anyone put an exact number on something so complex? The uncertainty seems far greater than a simple percentage.
I find it hard to believe such a specific timeline and probability. This feels like sensationalism rather than rigorous science.
Ten percent sounds low, but in existential risk terms, that is absolutely terrifying. We need to take this seriously.