“[On how much time is left before AI poses catastrophic risk:] We might have one year, two years, five years. I don't think we have 10 years.”
Co-founder and CEO of Conjecture; co-founder of EleutherAI
Said Jan 2024Deadline Jan 1, 2034TIME, 'To Stop AI Killing Us All, First Regulate Deepfakes, Says Researcher Connor Leahy', January 2024 (interview ahead of his Jan. 17 Davos fringe-event appearance)
Deadline is January 2034; as of 2026 no catastrophic AI outcome has occurred, but the claim's outer bound has not yet been reached.