OpenAI Expands Cybersecurity Efforts with GPT-5.5-Cyber and Patch the Planet
OpenAI has released GPT-5.5-Cyber, a specialized model for cybersecurity, alongside an update to the Codex Security plugin. The model is available only to verified users and is tuned for lower refusal rates on security-specific queries. The updated Codex Security scans codebases, analyzes attack vectors, checks reachability of vulnerable code, and generates patches, with support for external scanner bug reports and SARIF/CodeQL export.
In partnership with HackerOne and Trail of Bits, OpenAI also launched Patch the Planet, an initiative for automated protection of open-source projects. cURL, Python, and Go have already joined the program.