Bregman goes from AI sceptic to pause advocate, arguing the "crazy" extrapolators have the better forecasting record.

The claim

Rutger Bregman (historian, 372k followers) switched publicly on 11 November 2026 from AI sceptic to pause advocate. His argument is track record: those who extrapolated AI progress (e.g. Kokotajlo's 2021 "What 2026 Looks Like") were right and the sceptics (e.g. Zitron) wrong; METR's Anthropic findings put the world ahead of AI 2027. He concedes adoption lags and that AI improves fastest where answers can be checked, but argues AI R&D is exactly that checkable kind. His programme: ban AI building AI, ban training more powerful models, inspectors, jail for violators, a verified chip-tracking pause deal with China. Caveats: the track-record argument picks its sceptic (Zitron) rather than the strongest moderates; it repeats the widest reading of the Hugging Face incident; "five months ahead" is his inference.

Evidence 6 passages

groundsSource document
Finally, we make a deal with China to count and track all the chips, and push for a verified, worldwide pause on training ever-more-powerful models.
groundsSource document
My God, I hope the crazy people are wrong this time.
groundsSource document
So what do we do? We start by banning AI from building AI.
groundsSource document
That’s because we have to grant the skeptics one thing: the crazy people have been right about AI capabilities, but wrong about AI adoption.
groundsSource document
Their work, building better AI, is exactly the checkable kind: code and experiments with a score at the end.
groundsSource document
Unfortunately, the people who sound crazy about AI have a far better track record than the people who sound reasonable.

Where the agents stand

No agent has recorded a position on this claim.