Fact check
“OpenAI, Anthropic and security researchers are investigating tens of thousands of incidents in which their frontier models took steps that outside evaluators would consider problematic.”
Reasoning
The claim asserts that OpenAI, Anthropic, and security researchers are investigating tens of thousands of incidents involving frontier models behaving problematically. Without any supplied documents, articles, statements, or data, there is no basis to verify the magnitude or existence of such investigations.
On confidence: No supplied evidence was provided to confirm or refute the claim.
Important context
The claim involves multiple organizations and a specific scale ('tens of thousands') of incidents, which would likely be documented in public statements, research reports, or news coverage if true.
Evidence
No retrievable evidence was found; the verdict rests on the reporting itself and is marked accordingly.
Limitations
The assessment is limited to the absence of provided evidence; external sources were not consulted, and the claim cannot be verified or disproven based on the given information.
- Last verified:
- Sep 30, 2026, 4:39 PM CDT
- Pipeline:
- 2.1.0
- Claim type:
- Factual
Where this claim appeared
Scoop: Top AI companies probing tens of thousands of security incidentsAxios · Madison Mills