Ex-DeepMind safety intern David Krueger: alignment won't work, we must stop AI now

DavidSKrueger · x · 2026-09-10

David Krueger, an early intern on Google DeepMind's first AI safety team in 2018, published a long statement calling for an immediate halt to AI development. His reasons for leaving then: he was told not to mention YouTube in his research on manipulative algorithms, he didn't believe any AI company truly cared about safety, and he never trusted technical fixes. He argues nuanced governance is now too slow, that quitting researchers "made a bad bet" trusting companies or alignment timelines, and that "we must stop. There is no alternative."

Related event: Cambridge's Krueger Calls for Immediate Halt to Frontier AI Development(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →