AIAI securityStory 02
OpenAI reports 948 Linux privilege-escalation exploits and candidate fixes
What changedOpenAI reported on November 4 that a research pipeline produced 948 working local privilege-escalation exploits and candidate patches for Linux. The run took place in mid-October using an upcoming cyber model. OpenAI reports approximately $1.2 million in API costs, with human researchers reviewing outputs before reporting findings.
The useful part
Why it matters
Takeaway: Security teams need triage, deduplication, patch review and deployment capacity before scaled AI discovery can reduce real exposure.
Keep in mind
Good to know
Candidate patches are not equivalent to merged or deployed fixes, and the result should not be extrapolated to unrelated codebases.
Evidence