Jared kaplan, anthropic's key figure, warns of ai's self-training risks

Jared Kaplan, a key figure at Anthropic, has warned of the dangers of allowing AI to train itself, a concern he raised late last year. He predicted that this critical decision point could arrive between 2027 and 2030, a timeline that seems to be approaching as 2026 draws to a close. Kaplan, who co-founded Anthropic in 2021 with Dario Amodei, oversees AI safety and responsible scaling policies at the company. His warnings have gained traction, with around two-thirds of Americans now believing AI could pose an existential risk to humanity, according to a recent Politico poll. Kaplan's concerns have also reached policymakers, with lawmakers considering new legislation on AI safety. Jeffrey Ladish, a former Anthropic engineer, cited Kaplan's remarks during meetings with lawmakers, emphasizing the risks of AI self-training. Kaplan, now a billionaire, remains a respected leader at Anthropic despite the company's refusal to grant him an interview. His background in physics and contributions to AI research, including co-authoring a landmark paper on scaling laws, have positioned him as a pivotal figure in the field.