Announcing the Corrigibility Research Fund

The AI Alignment Forum reports that researcher Max Harms is launching a new fund focused on corrigibility research — the study of how to build AI systems that keep humans in control and amenable to correction. The fund, housed at Lightcone Infrastructure and backed by philanthropist Peter McCluskey, will award at least two hundred thousand dollars in grants and prizes during twenty twenty-six. Applications for the first grant round close on August twenty-third, with typical awards ranging from five thousand to thirty-five thousand dollars; a second round opens later in the fall. Corrigibility remains underfunded in AI safety despite backing from researchers like Eliezer Yudkowsky and Paul Christiano, and appears in Anthropic's latest constitution more than a dozen times.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton