OpenAI launches Daybreak, a security initiative to detect and patch code vulnerabilities

OpenAI announced Daybreak, an AI initiative that uses the Codex Security AI agent (launched in March) to create threat models based on an organization's code, identify possible attack paths, and automate detection of higher-risk vulnerabilities.

Daybreak combines OpenAI's most capable models, Codex, and security partners, and includes specialized cyber models: GPT-5.5 with Trusted Access for Cyber and GPT-5.5-Cyber, which began rolling out last week.

The launch follows Anthropic's announcement of Claude Mythos, a security-focused AI model that Anthropic said was too dangerous to publicly release and shared only privately as part of Project Glasswing.