AnalysisPolicySeptember 23, 2026

Anthropic's Jared Kaplan warned AI self-training could trigger intelligence explosion

Read original source →businessinsider.com

Kaplan said letting AI fully train itself could cause an "intelligence explosion" and called it the "ultimate risk," with a decision point possibly arriving between 2027 and 2030. As 2026 ends, state-of-the-art systems at Anthropic and OpenAI already help researchers build models.

People · Jared Kaplan

1 source

More stories today

Open the live feed