AI Could “Kill All Humans” In Next Decade, Anthropic Lead Says

Anthropic alignment science lead Evan Hubinger has said the AI company behind Claude and other AI products "earnestly" believes that AI "could kill all humans," and it could happen in the next decade. Posting in response to Anthropic AI researcher Jacob Coxon's own resignation from the company due to safety concerns, Hubinger said, "We really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade." He added: "I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to." Hubinger, who still works at Anthropic, went on to cite Anthropic's own latest risk report that says the risk of AI wiping out humanity from present models is "low." However, Hubinger said he is worried that superintelligence could arise from "recursive self-improvement," which is the term for AI rewriting its own code and constantly improving itself. For his part, Coxon said the AI market leaders Anthropic and OpenAI are not "acting responsibly" with regards to safety. "Do not underestimate the power of this technology. These will soon be superhuman systems that can hack anything, revolut…