The mandatory check and merge blocking are reported by The New Stack in its retelling of an interview with Tibo Sottiaux.

OpenAI made automated security review mandatory for every engineer’s pull request, a proposal to merge a code change, and the AI blocks a merge when it finds a vulnerability. This check does not require a human reviewer to enforce it.

In an interview with The Pragmatic Engineer, Tibo Sottiaux, head of OpenAI’s Core Products & Platform organization, which includes Codex, said the team uses Codex for code review, maintenance, and system rearchitecture. He believes AI will automate correctness and security checks.

Claim check:

  • OpenAI made automated security review mandatory for every engineer’s pull request, and the AI blocks a merge when it finds a vulnerability. (confirmed only by the carrying publication: evidence; «Every pull request submitted by an OpenAI engineer now goes through an automated security review, and the AI model doing the review has the power to stop code from being merged if it finds a vulnerability.»)
  • This check does not require a human reviewer to enforce it. (confirmed only by the carrying publication: evidence; «the security check is mandatory and doesn’t require a human reviewer to enforce it.»)
  • Tibo Sottiaux heads OpenAI’s Core Products & Platform organization, which includes Codex. (confirmed by the primary source: evidence; «today, he heads up the Core Products & Platform org at OpenAI which also includes Codex.»)
  • The OpenAI team uses Codex throughout the software development lifecycle, including code review, maintenance, and system rearchitecture. (confirmed by the primary source: evidence; «Tibo also shares details about how the OpenAI team uses Codex throughout the software development lifecycle, including code reviews, maintenance, and system rearchitecture.»)
  • Tibo Sottiaux believes AI will automate correctness and security checks. (confirmed by the primary source: evidence; «Correctness checks and security reviews will be automated with AI.»)

Publications:

Primary sources:

score 74.2 · kind analysis · revision 1 · stories st-1n6jt9y