OpenAI’s president and co-founder, Greg Brockman, has announced two new cyber products: Daybreak Red Team and Daybreak Defense Factory. He said both make it much easier to put frontier cyber capabilities to work, helping defend companies in new ways. The announcement followed OpenAI’s Cyber at the Frontier keynote.
Daybreak Red Team is described on its official page as AI penetration testing for web applications. OpenAI says you can put a coordinated team of agents to work testing your web applications, validating vulnerabilities, and checking if patches prevent the exploit.
The agents, the page says, use source code and app context to investigate attack paths and confirm vulnerabilities. They can also test findings from code scanners, bug-bounty submissions and disclosure reports, then use that evidence to help you decide what to fix first. After a patch, they can test whether it stops the exploit, and you can repeat the assessment as the apps change.
OpenAI says assessment scope is enforced with isolated environments and Guardian review, and that you control which apps and endpoints the agents can reach. Its Daybreak access guide says the programme is meant for lawful, authorised work on systems you own, operate, or are explicitly authorised to test, and that approval is reviewed rather than automatic.
Announcing two new cyber products: Daybreak Defense Factory and Daybreak Red Team. Both make it much easier to put frontier cyber capabilities to work, helping defend companies in new ways.
November 12, 2026
Adam, who works in go-to-market at OpenAI, shared the official Red Team page and quoted that product description.
AI penetration testing for web applications: Put a coordinated team of agents to work testing your web applications, validating vulnerabilities, and checking if patches prevent the exploit.
November 12, 2026
Daybreak Defense Factory was announced alongside it. Official pages say it can take findings from your existing security tools and from Red Team, then have agents investigate each issue and develop a patch. Engineers review and test proposed patches before merge, and can work on a fix with Codex while agents continue in the background. You can also scan new commits or schedule recurring checks as the code changes.
Trail of Bits, quoting Brockman’s announcement, said patching is the hardest step for agents to get right. The firm has released two open-source agent plugins from its Patch the Planet work and internal Codex sessions, and said they fit naturally in loops like OpenAI’s new Daybreak Defense Factory, where agents draft patches for people to review.
Patching is the hardest step for agents to get right. We took what we learned from hundreds of Patch the Planet patches, thousands of internal Codex sessions, and a decade of fix reviews, and built it into two open-source agent plugins.
November 12, 2026
Bug-bounty researcher Marius du Preez quoted the same announcement and wrote, “prove it on black box labs. no source, no hints.” Official pages say Red Team agents use source code and app context.
prove it on black box labs. no source, no hints
November 12, 2026
Tibo, who works on Codex and ChatGPT at OpenAI, said the company announced new tools it’s developing for cybersecurity, across automated patching and automated red teaming, and posted a link to watch the Cyber at the Frontier recording.
We announced new tools we’re developing for cybersecurity today across automated patching and automated red teaming. Watch the recording here
November 13, 2026
The same page points people to request Daybreak Access, and the official Red Team page invites you to start with an application you need to test.
Discover more from Sulat AI & Tech News
Subscribe to get the latest posts sent to your email.
