OpenAI

Cybersecurity · Artificial Intelligence

Launch

OpenAI launches Daybreak Red Team, Defense Factory cyber tools

Pitches security as a function of model quality, implying firms with weaker AI access become easier targets.

• OpenAI unveiled Daybreak Red Team and Daybreak Defense Factory at its Cyber at the Frontier keynote on November 12, 2026, expanding its Daybreak security platform with two new AI-agent products.
• Daybreak Red Team, now in early access, uses AI agents to analyze source code and app context, map attack paths, verify vulnerabilities from bug bounties or scanners, and test whether patches close them.
• Daybreak Defense Factory scans codebases and drafts merge-ready pull requests with fixes and tests, importing findings from tools like Wiz, Snyk and Endor Labs for engineers to review before deployment.
• Both products run on GPT-6 Cyber, OpenAI's newest cybersecurity-focused model, which the company says had just been used to find nearly 1,000 vulnerabilities in the Linux kernel.
• Daybreak launched in May 2026 and was already expanded in June with Codex Security updates and GPT-5.5-Cyber, so these tools extend an existing platform rather than start a new one.
• OpenAI and Anthropic already offered overlapping vulnerability-finding and patch-validation tools, leaving the real test of Daybreak's value in how much manual triage work it actually removes.
• The launch doubles as a sales pitch tying enterprise security to frontier model access, with OpenAI president Greg Brockman arguing companies are only as secure as the AI model defending them.