ClaimsDario Amodei

“10 to 25% referred to the chance of civilization getting substantially derailed, right?”

Dario Amodei, Anthropic CEO, interviewed 28 February 2025 Hard Fork - Anthropic's C.E.O. Dario Amodei on Surviving the A.I. Endgame Transcript (Podscripts (Hard Fork)) archived copy

Factual AI safety and alignment

Amodei's earlier 10 to 25% figure referred to the chance of civilization getting substantially derailed.

Resolved true The record shows it happened or was so.

Factual: Statements about the present or the past, checked against the record: did they hold up?

What would count as true or false

Judge whether Amodei's earlier public statements of a 10–25% figure described it as the chance of civilisation going badly or being substantially derailed (e.g. catastrophically wrong outcomes from AI). True: the earlier statements (e.g. his 2023 interviews or podcast remarks) describe the 10–25% figure in terms equivalent to civilisation being substantially derailed or things going catastrophically wrong. Mostly true: the figure was framed as the chance of things going 'really, really badly' or catastrophically, which is a close equivalent, with only minor wording differences. Mixed: the earlier statements are ambiguous, or different statements gave different framings. Mostly false: the figure referred to something materially narrower or different, such as human extinction specifically, though related. False: no 10–25% figure was given, or it referred to an unrelated quantity.

Set before the research began.

Evidence (4)

  1. E1 AI Kills Us All (with Daniel H. Wilson) Newcomer · 17 October 2023

    “Anthropic CEO Dario Amodei recently pegged his “chance that something goes really quite catastrophically wrong on the scale of human civilization” between 10% and 25%.”

  2. E2 Dario Said — Dario Amodei on AI, in his own words Dario Said (independent archive)

    “On the Logan Bartlett Show he put the chance of something going "quite catastrophically wrong on the scale of human civilization" at "somewhere between 10 and 25%."”

  3. E3 Dario Amodei: 10 to 25 percent chance that things go badly GAGE (Global Academy of Generative-AI Education), Doom Number ledger · 16 September 2026

    “spends much of his effort reducing the 10 to 25 percent chance that disaster could occur.”

  4. E4 Amodei on AI: "There's a 25% chance that things go really, ... Axios · 17 September 2025

    “Amodei gave the percentage after being asked his "(p)doom number," which is the probability that things go south with AI — specifically that it destroys humanity.”

Reasoning

A 2023 newsletter (E1) reports that Amodei put his "chance that something goes really quite catastrophically wrong on the scale of human civilization" at 10% to 25%. An independent archive (E2) gives the same wording for his Logan Bartlett Show remarks. The criteria list "things going catastrophically wrong" at civilisation scale as equivalent to "substantially derailed," so this matches the true condition. E3 describes the same range as a chance of disaster. The extinction framing in E4 comes from Axios's description of the p(doom) question in a later 2025 report, not from the earlier statements.

Cite this claim

Dario Amodei, “Amodei's earlier 10 to 25% figure referred to the chance of civilization getting substantially derailed.” (said 28 February 2025). Verdict: resolved true. Reliability API, claim c_92ef44f3cf61, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_92ef44f3cf61

This claim is in dataset release v1.1.2, with the same verdict.

Reply to this claim

For Dario Amodei, or someone speaking for them: a correction or counter-evidence. Replies are read before anything is shown; an approved reply shows your name, role, date, statement and link. Your email address is never shown and is deleted 30 days after the decision (privacy).

More factual claims by Dario Amodei

  1. Mostly true Factual 28 February 2025
    “So we did this experiment where we basically train the model to be all the good thing, helpful, honest, harmless, friendly.”

    Anthropic ran an experiment training a model to be helpful, honest, harmless and friendly.

  2. Resolved true Factual 28 February 2025
    “They're not yet at the stage where we think that there is a real and meaningful increase in the threat end to end,”

    Anthropic assessed that Claude 3.7 Sonnet is not yet at a stage giving a real and meaningful end-to-end increase in dangerous threat capability.

  3. Mostly true Factual 18 February 2025
    “this lack of understanding is essentially unprecedented in the history of technology.”

    The lack of understanding of how AI systems work is essentially unprecedented in the history of technology.

  4. Mostly true Factual 18 February 2025
    “We were able to provide a strong theoretical basis for having conviction that superposition was the right hypothesis to pursue.”

    Amodei said that "we" were able to provide a strong theoretical basis for having conviction that superposition was the right hypothesis to pursue; the surrounding text does not say who "we" refers to.

  5. Mostly true Factual 18 February 2025
    “What changed at this time was that it became clear this was going to be a central problem for language models, much worse than in vision.”

    It became clear that superposition would be a central problem for language models, much worse than in vision models.

All of them

Verdicts are made automatically; how verdicts are made.