OpenAI has introduced two cybersecurity products under the Daybreak name. Daybreak Red Team uses agents to inspect source code and application context, map possible attack paths and test web applications for vulnerabilities. It can also verify findings from scanners or bug-bounty reports and test whether a proposed patch closes the issue.
Red Team is available in early access, so OpenAI’s claim that it combines continuous automated scanning with the depth of a human penetration test has not yet been established across broad independent deployments. Security teams will also need to define safe targets and isolate testing from systems where an agent’s probes could cause harm.
Daybreak Defense Factory handles the remediation side. Its agents scan repositories, generate patches and include tests in pull requests that developers can review before deployment. The workflow keeps approval with a person rather than directly changing production code.
Both products run on GPT-6 Cyber, according to the announcement. Together they place OpenAI in a security market already filled with code scanners, managed testing services and AI-assisted repair tools. Their practical value will depend on false-positive rates, the quality of generated fixes and whether the agents can uncover multi-step flaws without introducing new defects.