Paul Christiano, a member of the OpenAI Foundation’s board and its Safety and Security Committee, has warned that there is a meaningful risk of an AI takeover in the next six to 12 months. Speaking at the Hinton Lectures 2026, hosted by Geoffrey Hinton, Christiano told the audience that for most people in the room, the most likely reason they will die is an AI accident, Gizmodo and Ars Technica reported.
His numbers come with his own caveat. Christiano said that if AI labs build as quickly as they can without intervention, there would be something like a coin flip chance of an AI takeover within three years. In September he put his all things considered estimate of a catastrophic loss of human control at 4 percent within one year and 15 percent within three years, and acknowledged those were subjective judgments rather than the product of a precise model.
Christiano co-invented reinforcement learning from human feedback, founded the Alignment Research Center, and joined OpenAI’s board in September specifically to help oversee its safety practices. That position is what gives the warning its weight in the governance debate: the committee he sits on holds authority over model releases under OpenAI’s agreement with California’s attorney general.
Other forecasters moved the same way this week. Eli Lifland of the AI Futures Project wrote that after reviewing new research suggesting AI is accelerating its own development, he worried humanity might no longer be in control of its future within 3 to 12 months. The claims remain forecasts by named researchers, not measurements, and no lab has announced a change to its release plans in response.