OpenAI just released its answer to Claude Mythos
The Verge
OpenAI has launched Daybreak, a new AI initiative designed to automate the detection and patching of software vulnerabilities.Summary
OpenAI has introduced Daybreak, an AI initiative aimed at identifying and remediating security vulnerabilities before they can be exploited. The system utilizes the Codex Security AI agent to build threat models, validate potential risks, and automate detection efforts. This move serves as a response to Anthropic's security-focused Claude Mythos and Project Glasswing. Daybreak integrates various OpenAI models, including specialized cyber-security versions of GPT-5.5, and is being developed in collaboration with government and industry partners to improve proactive cyber defense.
(Source：The Verge)