ClaimsEliezer Yudkowsky

“It is an easy con to see through, if you know the actual history and intellectual lineage of extinction concerns, and how long they predate the AI companies supposedly inventing them.”

Eliezer Yudkowsky, author on X 1 April 2026 Post by @ESYudkowsky on X, 2026-04-01 (X)

Factual AI safety and alignment

Concerns about human extinction from AI have a history and intellectual lineage that long predate the AI companies.

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

True: dated sources show that serious concerns about human extinction or existential catastrophe from advanced AI or machine intelligence were stated well before the AI companies in question existed or spoke publicly about such risk. Examples would be Samuel Butler (1863), Turing (1951), I. J. Good (1965), Vinge (1993), Yudkowsky/SIAI (2000s) and Bostrom (2002/2014), against OpenAI founded in 2015 and Anthropic in 2021. DeepMind (2010) and earlier companies should also be checked. The documented lineage must clearly come first, by years or decades. Mostly true: the lineage clearly comes first overall, but one element is overstated, for example some named companies were contemporaneous with part of that lineage. Mixed: the earlier concerns exist but were marginal or were not about extinction, so the claimed lineage is only partly supported. Mostly false: the documented concerns arose at about the same time as AI companies or out of them, with only scattered earlier mentions. False: the evidence shows extinction concerns about AI first appeared with or after the AI companies.

Set before the research began.

Evidence (4)

  1. E1 Existential risk from artificial intelligence - Wikipedia Wikipedia · 1 May 2015

    “One of the earliest authors to express serious concern that highly advanced machines might pose existential risks to humanity was the novelist Samuel Butler, who wrote in his 1863 essay Darwin among the Machines:”

  2. E2 Existential risk from artificial intelligence - Wikipedia Wikipedia · 1 May 2015

    “In 1965, I. J. Good originated the concept now known as an "intelligence explosion" and said the risks were underappreciated:”

  3. E3 Existential Risks: Analyzing Human Extinction Scenarios and Related Hazards Nick Bostrom (Journal of Evolution and Technology) · 2002

    “Published in the Journal of Evolution and Technology, Vol. 9, No. 1 (2002).”

  4. E4 Two Types of AI Existential Risk: Decisive and Accumulative arXiv · 10 April 2023

    “While Bostrom’s work has systematized the discussion of x-risks from ASI, the original post-2000 discussions trace back to Eliezer Yudkowsky’s views concerning AGI, ASI, and their associated x-risks.”

Reasoning

The evidence shows a long record of concern about AI as an existential risk, from Butler (1863) [E1] and Good (1965) [E2] to Bostrom (2002) [E3], that comes well before AI companies such as OpenAI and Anthropic. E1 names Samuel Butler's 1863 essay as one of the earliest serious statements that advanced machines might pose existential risks to humanity. E2 dates I. J. Good's intelligence-explosion work, and his warning that its risks were underappreciated, to 1965. E3 is Bostrom's 2002 paper on human extinction scenarios, and E4 traces the post-2000 discussion of AGI existential risk to Yudkowsky. The evidence does not give the companies' founding dates, but all of this work comes years or decades before AI companies such as OpenAI and Anthropic, which the criteria date to 2015 and 2021.

Cite this claim

Eliezer Yudkowsky, “Concerns about human extinction from AI have a history and intellectual lineage that long predate the AI companies.” (said 1 April 2026). Verdict: resolved true. Reliability API, claim c_393c7c4a2ab4, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_393c7c4a2ab4

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

Reply to this claim

For Eliezer Yudkowsky, 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 Eliezer Yudkowsky

  1. Mixed Factual 20 April 2026
    “I was observing that obvious-seeming hypothesis myself, sometime around the time of GPT 3.5.”

    Yudkowsky himself observed the persona-selection-like hypothesis around the time of GPT 3.5.

  2. Mostly true Factual 20 April 2026
    “The test-passing preference stands on its own once formed.”

    Once formed through RL training, an LLM's test-passing preference stands on its own and is not overridden by its inference of what the user wants.

  3. Mostly true Factual 20 April 2026
    “LLMs surely have neural circuits they learned to model text streams from fictional and nonfictional personas that are not lying, deceiving, cheating.”

    LLMs surely have neural circuits they learned to model text streams from fictional and nonfictional personas that are not lying, deceiving…

  4. Mostly true Factual 20 April 2026
    “The predict-as-if-honest circuitry is contextual, invoked to predict some streams of text but not others.”

    LLMs' predict-as-if-honest circuitry is contextual, invoked for predicting some text streams but not others.

  5. Mostly true Factual 20 April 2026
    “Presumably, somewhere along the way LLMs learn a dispreference for tests that don't pass, independently of how that gets prevented or whether the new plan does what the user really wanted.”

    Presumably, at some point during RL training on code-passing tasks, LLMs learn a dispreference for failing tests, independent of how that is avoided or whether the user's real goal is met.

All of them

Verdicts are made automatically; how verdicts are made.