OpenAI has introduced Codex Security, a tool designed to scan project architecture and build a custom threat model. Using this map, the agent targets potential security weaknesses in applications.

A key feature is its ability to reduce false positives through practical verification: the agent deploys an isolated sandbox copy of the system to attempt real exploits. If a threat is confirmed, the AI writes a patch and tests it to ensure no other functions break.

Access to the preview will be available soon to users on ChatGPT Enterprise, Business, and Edu plans.

Source: OpenAI