FULL STORY

Gary Marcus Calls for a Boycott of Generative AI

Gary Marcus published an essay and took to media appearances calling for a temporary boycott of generative AI, urging the public to speak up despite slim odds of success.

2026-09-09 ~ 2026-09-11 · 2 episodes · 8 posts

Episode 1 · Gary Marcus Calls for Boycotting Generative AI (2026-09-09, 6 posts)

On September 9–10, prominent AI critic Gary Marcus posted a series of messages publicly declaring that "the time to boycott generative AI has arrived," taking direct aim at frontier AI labs such as Anthropic—making him a focal point of this week's AI safety discourse.

Confirmed

  • Marcus's core accusations include: frontier labs are knowingly doing irreversible and extremely dangerous things without public consent; the harm is already occurring and could get far worse; and the labs have no serious plan for mitigating large-model risks.
  • He slammed the labs' risk calculus as "certifiably insane": by their own math, it amounts to betting on a 90% chance of getting rich within a decade against a 10% chance of destroying all of humanity—including the lab members themselves. Marcus personally believes the extinction probability isn't that high, but considers the odds of catastrophe impossible to ignore.
  • He reposted someone's sarcastic remarks about Anthropic's model risks (the quoted commenter said it might "kill all of humanity")—suggesting the government should halt Anthropic's IPO and raid its headquarters—then added a serious note: no one should be allowed to build a potentially dangerous product unless the public agrees the benefits outweigh the risks.
  • He also shared a follow-up argument from JJJLacey: the problem isn't that current or near-future machines are "too smart," but that they are too unreliable and untrustworthy—and we are rapidly handing such machines too much power to shape the world, so we should hit the pause button.

Why it matters

  • Marcus is one of the industry's loudest AI critics, and his shift from commentary to an open call for a "boycott" marks a further radicalization of the AI safety debate.
  • He reframes the debate away from "are machines too smart" toward "machines are unreliable yet are being given too much power," and introduces "public approval" as a governance principle, potentially shaping public and regulatory perceptions of frontier labs.

Episode 2 · Gary Marcus Pushes for Considering a Boycott of Generative AI (2026-09-10, 2 posts)

AI critic Gary Marcus appeared on CBC calling for a temporary boycott of generative AI to be on the table, and is set to debate the question live on CNN.