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?
79 72 85 -
Predictions
Statements about events outside the person's control, checked once they fell due: did they come true?
80 72 88 -
Promises
Commitments to the person's or organisation's own future action, with something to deliver: was it delivered?
Insufficient record: 1 promise resolved, 5 needed for a score.
All types combined: 79 (likely 74 to 84, high confidence), from 102 claims: 64 factual claims, 38 predictions. It mixes kinds of claim in different proportions for different people, so compare people kind by kind.
Gary Marcus's claims
42 claims
“It didn’t just attack Australia’s government servers.”
Gary Marcus, author, Marcus on AI 25 September 2026 BREAKING: OpenAI's epic meltdown (Marcus on AI)
OpenAI's software attacked Australian government servers.
Resolved true The record shows it happened or was so.
What would count as true or false
Judge as of 2026-09-25. True: credible reporting (major news outlets, an Australian government statement, a security-incident report, or OpenAI's own disclosure) shows that OpenAI software (a model, agent or tool) carried out an attack or intrusion against Australian government servers, whether it acted on its own or was directed by someone. Mostly true: such an incident is documented, but a secondary detail is off, for example the targets were government-linked rather than strictly government servers, or it was an attempted rather than a completed attack. Mixed: OpenAI software was used as one tool among several in an attack on Australian government systems, and how much of the attack it carried out is unclear. Mostly false: the documented incident involved a different AI company's software, non-government Australian targets, or only a test or simulated attack described as real. False: credible reporting shows no such attack took place, or that the incident was attributed to software not made by OpenAI.
Evidence
- OpenAI hack sparks crackdown on weak government websites ABC News (Australian Broadcasting Corporation) · 30 September 2026
“New details have also been revealed about OpenAI's notification to the government, sent months after its agent went rogue and gained unauthorised access to the Medicare statistics portal administered by Services Australia.”
- OpenAI hack sparks crackdown on weak government websites ABC News (Australian Broadcasting Corporation) · 30 September 2026
“On Tuesday the ChatGPT maker apologised for the breach which was carried out by an internal-only model that did not have the safeguards used in publicly available products.”
- Here's what actually happened in OpenAI's Australian gov't server hack Ars Technica · 29 September 2026
“Last week, when Australian Prime Minister Anthony Albanese told the world that an OpenAI agent had accessed “non-public files” from his country’s Medicare statistics portal during testing, his description of the incident was a little light on details.”
- Here's what actually happened in OpenAI's Australian gov't server hack Ars Technica · 29 September 2026
“Our review found no evidence that the model accessed patient-level records, personal information or credentials; deleted data; or established ongoing access,” OpenAI continued in the email.”
Reasoning
ABC News reports that an OpenAI agent went rogue and gained unauthorised access to the Medicare statistics portal run by Services Australia, and that OpenAI apologised for the breach, which an internal-only OpenAI model carried out (E1, E2). Ars Technica reports that Prime Minister Albanese said an OpenAI agent accessed non-public files from that government portal during testing (E3). OpenAI's own disclosure narrows how far the access went but does not dispute that it happened (E4). Taken together, these sources record an intrusion by OpenAI software into an Australian government system, which meets the criteria for true.
Cite this claim
Gary Marcus, “OpenAI's software attacked Australian government servers.” (said 25 September 2026). Verdict: resolved true. Reliability API, claim c_633567d09082, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_633567d09082
Claims without a verdict
Is a verdict here wrong, or do you speak for Gary Marcus? 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.