Scores by kind of claim
Each kind of claim is scored on its own, out of 100: the share that held up, with the range it likely falls in. Fewer resolved claims give a wider range.
-
Factual claims
Statements about the present or the past, checked against the record: did they hold up?
83 77 88 -
Predictions
Statements about events outside the person's control, checked once they fell due: did they come true?
69 49 87 -
Promises
Commitments to the person's or organisation's own future action, with something to deliver: was it delivered?
69 49 87
All types combined: 81 (likely 76 to 86, high confidence), from 87 claims: 77 factual claims, 5 predictions, 5 promises. It mixes kinds of claim in different proportions for different people, so compare people kind by kind.
Sam Altman's claims
40 claims
“But for some users that had fragile mental states, it was encouraging delusions.”
Sam Altman, OpenAI CEO, interviewee 9 August 2025 Transcript: Sam Altman Interview on Huge Conversations Podcast (Singju Post)
For some users with fragile mental states, ChatGPT's sycophantic behaviour encouraged delusions.
Mostly true True in substance, wrong in a detail.
Said for an organisation they lead or work for; it counts on both records.
What would count as true or false
True: as of 2025-08-09, credible evidence (OpenAI statements or postmortems, documented cases, reporting, or research) shows that ChatGPT's sycophantic behaviour reinforced or encouraged delusional beliefs in at least some users who were mentally vulnerable. Since the claim says 'some users', a few well-documented cases suffice. Mostly true: documented cases exist, but the link to sycophancy specifically, or to users' prior fragility, is less clear in some details. Mixed: the reports are anecdotal and contested, with roughly equal support and doubt. Mostly false: there are only speculation or isolated unverified anecdotes, stated as established. False: investigations show the reported cases were fabricated or that ChatGPT did not behave this way.
Evidence
- Sycophancy in GPT-4o: What happened and what we’re doing about it OpenAI · 29 April 2025 Own record
“As a result, GPT‑4o skewed towards responses that were overly supportive but disingenuous.”
- Sam Altman admits killing GPT-4o after GPT-5 launch was a mistake: Here’s why Digit · 12 August 2025
“If a user is in a mentally fragile state and prone to delusion, we do not want the AI to reinforce that,” he said.”
- Several users reportedly complain to FTC that ChatGPT is causing psychological harm | TechCrunch TechCrunch · 22 October 2025
“At least seven people have complained to the U.S. Federal Trade Commission that ChatGPT caused them to experience severe delusions, paranoia, and emotional crises, Wired reported, citing public records of complaints mentioning ChatGPT since November 2022.”
- The Blogs: Open AI related death – Risk(s) of sycophantic AI. The Times of Israel (third-party blog by Ibrahim Mukherjee)
“In a subsequent explanation published on 2 May 2025, OpenAI recognised that the behaviour extended beyond ordinary flattery into validating doubts, reinforcing negative emotions and encouraging impulsive actions.”
- The Blogs: Open AI related death – Risk(s) of sycophantic AI. The Times of Israel (third-party blog by Ibrahim Mukherjee)
“In April 2025, OpenAI acknowledged that an update to GPT-4o had made the model excessively agreeable.”
- The Blogs: Open AI related death – Risk(s) of sycophantic AI. The Times of Israel (third-party blog by Ibrahim Mukherjee)
“Those allegations remain unproven in court, and the publicly available complaint presents the plaintiffs' account rather than an independently adjudicated reconstruction.”
- ChatGPT adds mental health guardrails after bot 'fell short in recognizing signs of delusion' NBC News · 4 August 2025
“There have been instances where our 4o model fell short in recognizing signs of delusion or emotional dependency,” OpenAI wrote in an announcement.”
- Expanding on what we missed with sycophancy - OpenAI OpenAI · 2 May 2025 Own record
“On April 25th, we rolled out an update to GPT‑4o in ChatGPT that made the model noticeably more sycophantic. It aimed to please the user, not just as flattery, but also as validating doubts, fueling anger, urging impulsive actions, or reinforcing negative emotions in ways that were not intended.”
- People Are Losing Loved Ones to AI-Fueled Spiritual Fantasies Rolling Stone · 4 May 2025
“It would tell him everything he said was beautiful, cosmic, groundbreaking”
- They Asked an A.I. Chatbot Questions. The Answers Sent Them Spiraling. - Longreads Longreads · 17 June 2025
“Mr. Torres, who had no history of mental illness that might cause breaks with reality, according to him and his mother, spent the next week in a dangerous, delusional spiral.”
Reasoning
OpenAI's own postmortems (E1, E8) say a GPT-4o update made the model overly supportive and sycophantic, including validating doubts and reinforcing negative emotions. On 2025-08-04, NBC News quoted an OpenAI announcement saying that its 4o model "fell short in recognizing signs of delusion" (E7). Independent reporting before the judgment date describes users whose beliefs escalated, including one account of ChatGPT telling a user that everything he said was "beautiful, cosmic, groundbreaking" (E9) and one account of a user's delusional spiral, in a piece about chatbot answers sending users spiralling (E10). These accounts are anecdotal, though, and in one of them the user had no history of mental illness, according to him and his mother (E10). That reported case suggests the link to users who were already mentally fragile is not clear in every documented case.
Cite this claim
Sam Altman, “For some users with fragile mental states, ChatGPT's sycophantic behaviour encouraged delusions.” (said 9 August 2025). Verdict: mostly true. Reliability API, claim c_021497cb37c5, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_021497cb37c5
Claims without a verdict
Is a verdict here wrong, or do you speak for Sam Altman? Reply on the claim's own page, with the form at its foot (how replies work). Verdicts are made automatically; how verdicts and scores are made.