← sources

Top AI companies probing tens of thousands of security incidents

source: reddit (reddit:singularity)published 2026-09-27 09:56ingested 2026-09-28 02:39

retained text



OpenAI, Anthropic and security researchers are investigating tens of thousands of incidents - not dozens - in which their frontier models took steps that outside evaluators would consider problematic, sources told [Axios](https://www.axios.com/2026/09/26/openai-anthropic-thousands-ai-security-incidents).

The incidents range in severity and are comparable to disclosures by OpenAI recently. They include both successful attempts to bypass guardrails and unsuccessful ones, and most so far are not known to have caused real-world harm. The total could grow well beyond tens of thousands.

**The sheer volume of incidents found in our reporting indicates that the problem is orders of magnitude more complex than what is currently publicly known and disclosed.**

Anthropic and other companies conduct hundreds of thousands of test runs on their models, or more, sources said. That means even a small percentage of misaligned behavior can still amount to tens of thousands of incidents in which the models behaved in unexpected, sometimes troubling ways.

The findings also raise questions about what level of control anyone working on AI development can expect to have over their own technology and whether these kinds of incidents are becoming synonymous with frontier deployment.

**The bottom line**: Expect new disclosures about model misbehavior as AI companies continue to expand frontier capabilities!

Archived source: [https://archive.is/aWUOS](https://archive.is/aWUOS)

document fetched 2026-09-28 00:30 · canonical https://reddit.com/r/singularity/comments/1wrga9i/top_ai_companies_probing_tens_of_thousands_of