ClaimsEliezer Yudkowsky

“even if GPT-5 doesn’t end the world, which I agree is like more than 50% of where my probability mass lies”

Eliezer Yudkowsky, interviewee 6 April 2023 Eliezer Yudkowsky - Why AI Will Kill Us, Aligning LLMs, Nature of Intelligence, SciFi, & Rationality (Dwarkesh Podcast) archived copy

Prediction AI safety and alignment

GPT-5 will not end the world, with more than 50% of Yudkowsky's probability on that outcome.

Resolved true The record shows it happened or was so.

Prediction: Statements about events outside the person's control, checked once they fell due: did they come true?

The second judge would not resolve this claim; the verdict shown is the first judge's.

What would count as true or false

This is a forecast, stated as more than 50% likely, that GPT-5 will not end the world. True: OpenAI released a model officially named GPT-5, and no extinction-level or civilization-ending event attributable to it has occurred. Mostly true: GPT-5 was released and the world did not end, but GPT-5 was linked to serious, non-civilizational harms. Mixed: does not normally apply. Mostly false: GPT-5 contributed to a global catastrophe short of ending the world. False: GPT-5 ended the world or caused a civilization-ending catastrophe. If no model named GPT-5 has been released, the claim is unresolvable.

Set before the research began.

Evidence (4)

  1. E1 Introducing GPT-5 OpenAI · 7 August 2025

    “GPT‑5 is starting to roll out today to all Plus, Pro, Team, and Free users, with access for Enterprise and Edu coming next week.”

  2. E2 New warnings about the risks of AI to humanity revive a long-running debate WTOP News (AP) · 14 September 2026

    “There is no widely accepted estimate for how soon any of these scenarios might happen and no consensus on their likelihood.”

  3. E3 New warnings about the risks of AI to humanity revive a long-running debate WTOP News (AP) · 14 September 2026

    “OpenAI described the intrusion by a combination of models, including its newly released GPT‑5.6 Sol and an “even more capable” model that was still being tested internally, as a “significant security incident.””

  4. E4 Seattle woman’s family sues OpenAI, claims ChatGPT coached her off medication before death KIRO 7 · 29 October 2026

    “OpenAI spokesman Drew Pusateri told The New York Times that Wan’s story was an “incredibly heartbreaking situation,” but said Wan was using an early version of ChatGPT that is no longer available.”

Reasoning

E1 is OpenAI's official announcement, and it shows that a model named GPT-5 began rolling out on 2025-08-07. News reports dated September and October 2026 (E2, E3, E4) show that ordinary life and public debate were still going on more than a year later, so no civilization-ending event had happened. None of the evidence ties serious harm to GPT-5 itself: the security incident in E3 involves GPT-5.6 Sol and an unreleased model, and the lawsuit in E4 concerns "an early version of ChatGPT" that the report does not identify as GPT-5. The forecast that GPT-5 would not end the world therefore resolves true.

Cite this claim

Eliezer Yudkowsky, “GPT-5 will not end the world, with more than 50% of Yudkowsky's probability on that outcome.” (said 6 April 2023). Verdict: resolved true. Reliability API, claim c_9ff3db86aa71, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_9ff3db86aa71

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 prediction claims by Eliezer Yudkowsky

  1. Resolved false Prediction 6 April 2023
    “The current crop of people will not try it.”

    The current AI developers will not try the alignment approach he described (filtering training data to nice examples and training on it).

  2. Resolved true Prediction 6 April 2023
    “Alignment will not be solved in a few years.”

    AI alignment will not be solved within a few years.

  3. Mostly true Prediction 2008
    “Most AI researchers will not want to make unFriendly AIs.”

    Most AI researchers will not want to make unFriendly AIs.

  4. Mixed Prediction 5 March 2024
    “Beating GPT-4 on GI just forces OpenAI to release 4.5.”

    If a rival AI developer beats GPT-4 on general intelligence, OpenAI will be forced to release GPT-4.5.

  5. Mostly false Prediction 6 April 2023
    “So probably Paul’s going to win, but slight moral victory.”

    Paul Christiano will probably win his forecast disagreement with Yudkowsky on an AI getting IMO gold by 2025, with prediction-market odds then running around 30%.

All of them

Verdicts are made automatically; how verdicts are made.