What happened
OpenAI expanded its Daybreak initiative, launching an updated Codex Security plugin and the full GPT-5.5-Cyber model. The Codex Security plugin, which scanned over 30 million commits across 30,000 codebases, now accelerates vulnerability discovery and patching, preventing new vulnerabilities from reaching production. GPT-5.5-Cyber, available through limited release to trusted defenders, achieved 85.6% performance on CyberGym, an increase from GPT-5.5's 81.8%. OpenAI also introduced the Daybreak Cyber Partner Program and the "Patch the Planet" initiative with Trail of Bits, HackerOne, and Calif, involving over 30 open-source projects including cURL, Go, Python, Sigstore, and pyca/cryptography.
Why it matters
AI models now accelerate vulnerability discovery, shifting the cybersecurity bottleneck from finding to patching. Security architects and open-source maintainers gain access to tools that navigate large codebases, identify issues, and generate targeted patches at scale. This expansion, following Daybreak's initial launch in May, democratises advanced defensive capabilities, enabling teams to fix vulnerabilities before attackers exploit them, as evidenced by Codex Security's 500,000 automatically fixed findings.
