Codex Security Cloud adds continuous repository scanning

OpenAI is turning application security from periodic scanning into a persistent, context-aware engineering workflow. The approach is compelling, but teams should treat it as an investigative layer alongside deterministic scanners, not a replacement.

• –Continuous commit monitoring keeps security analysis closer to the point where vulnerabilities are introduced.
• –Repository-specific threat models, isolated validation, and deduplication aim to reduce noisy false positives.
• –Proposed patches still require human review; Codex Security does not automatically modify production code.
• –Daybreak Blue broadens model access for authorized defensive work, but its inclusion is limited to Codex Security Cloud and does not extend to the API or other products.
• –Research-preview availability, token-based billing, and lengthy scans for large repositories will matter for enterprise rollout.