Gary Marcus Calls for a Public Boycott of Generative AI
From September 9 to 11, prominent AI critic Gary Marcus posted a series of messages and published a Substack essay, "The Case for Boycotting Generative AI," declaring that "the time for boycotting generative AI has come," aiming squarely at frontier AI labs such as Anthropic. The move escalates his long-running criticism into an explicit boycott campaign and became a focal point of this week's AI safety discourse.
Confirmed
- Marcus's core charges: frontier labs are knowingly and admittedly doing irreversible, extremely dangerous things without public consent; harm is already occurring and could worsen dramatically; and the labs have no serious mitigation plans for large-model risks.
- He called the labs' risk calculus "certifiably insane": by their own math, a 90% chance of vast wealth within a decade is being wagered against a 10% chance of destroying all of humanity, including the labs' own people. Marcus personally thinks the extinction probability is lower, but says catastrophic risk is non-negligible; he also calls literal existential risk (x-risk) a distraction.
- After amplifying a joke about Anthropic's model risks (the quoted poster said it could "kill all of humanity" and suggested the government halt Anthropic's IPO and raid its headquarters), Marcus turned serious: this is not a joke, and no one should be allowed to build potentially dangerous products unless the public agrees the benefits outweigh the risks.
- He also amplified JJJLacey's argument that the problem is not that current or near-future machines are too smart, but that they are too unreliable and untrustworthy while being rapidly given too much power to influence the world—hence the call to pause.
- Replying to a user, Marcus said the current strategy is to use social pressure to stop AI labs from becoming richer and more powerful, reducing the chance of catastrophic harm.
- On September 11, Marcus consolidated the thread of arguments into the Substack essay "The Case for Boycotting Generative AI," formalizing the call to action.
Why it matters
- Marcus is one of the most audible AI critics in the field; moving from commentary to an explicit, essay-backed boycott call marks a further radicalization of the AI safety debate.
- He reframes the debate from "are machines too smart" to "machines are unreliable yet being granted too much power," and introduces "public approval" as a governance demand, potentially shaping public and regulatory attitudes toward frontier labs.
- Notably, he denies endorsing high extinction-probability narratives and calls x-risk a red herring, yet still calls for a boycott on catastrophic-risk grounds—a tension that itself is a useful lens on where AI safety discourse is heading.
2026-09-09 ~ 2026-09-11 · 8 related posts
Primary sources
- Gary Marcus calls for boycotting Generative AI, citing Anthropic and OpenAI scientists' own admissions — GaryMarcus ·
- Gary Marcus calls for a public boycott of generative AI: no risk plan, time to act — GaryMarcus ·
- Gary Marcus: no one should build potentially dangerous AI without public sign-off — GaryMarcus ·
- [source] Gary Marcus: no one should build potentially dangerous AI without public sign-off — GaryMarcus · 2026-09-09
- Gary Marcus calls for public boycott of frontier AI labs, citing Anthropic researcher's >10% doom estimate — GaryMarcus · 2026-09-09
- Gary Marcus: frontier labs bet 90% odds of riches against 10% odds everyone dies — GaryMarcus · 2026-09-09
- [source] Gary Marcus calls for boycotting Generative AI, citing Anthropic and OpenAI scientists' own admissions — GaryMarcus · 2026-09-10
- Gary Marcus makes the case for boycotting generative AI over reliability — GaryMarcus · 2026-09-10
- Gary Marcus: strategy is community pressure to stop AI labs getting richer and more powerful — GaryMarcus · 2026-09-11
2 near-duplicate retellings: GaryMarcus · GaryMarcus