OpenAI launched two new cybersecurity products inside its Daybreak platform on November 12, 2026: Daybreak Red Team and Daybreak Defense Factory. The announcement came during the Cyber at the Frontier keynote, delivered by company president Greg Brockman. The idea is to bring frontier AI to defense: agents that test applications for flaws, and others that turn findings into fixes ready for human review.
Daybreak Red Team: continuous pentesting with agents
Daybreak Red Team is an AI-agent penetration-testing service. Instead of a one-off scan, it continuously probes web applications: the agents analyze source code and application context, test attack paths, and check whether a flaw can actually be exploited — and whether a patch stops the exploit. Nancy Wang, who leads cyber-applied AI at OpenAI, described the service as superhuman penetration testing at the scale of automated, continuous scanning.
The product is in early access with selected customers, and a waitlist is open. OpenAI did not disclose pricing.
Daybreak Defense Factory: from finding to fix
Daybreak Defense Factory closes the other side of the loop: agents built on OpenAI's latest cybersecurity models — currently GPT-6 Cyber — scan codebases, find and validate vulnerabilities, and prepare patches for engineers to review. Nothing is applied automatically: the final decision stays with a person.
The tool is available in research preview through the Daybreak plugin in Codex. A GitHub repository can be connected for a one-off scan or for continuous monitoring of new commits, with review of commit history and the project's threat model. According to OpenAI, Codex Security Cloud has already analyzed more than 30 million commits across more than 30,000 codebases.
Brockman's thesis
In the keynote, Brockman said companies will now be secure only to the extent of the frontier model they use to defend themselves — and how well they use it. The statement sums up OpenAI's bet: as agents become able to run ever-longer cyber operations, defense must reorganize around the most capable models, without abandoning security fundamentals.
$1 billion for frontline defenders
On the same day, OpenAI also announced Daybreak for Frontline Defenders, a global $1 billion commitment in subsidized access to Daybreak's cybersecurity capabilities, training, and technical support, targeted to be consumed over the next six months. The focus is resource-constrained defenders protecting essential services — water, electricity, local government, and community banks — starting in the United States. The initiative includes a pilot with the Multi-State Information Sharing and Analysis Center (MS-ISAC) and more than 35 partner products and services in the Daybreak Defense Network. According to the company, thousands of defenders across 2,000 approved organizations already use Daybreak.
Not to be confused: Red Team is not Daybreak Red
Daybreak Red Team is a new product and should not be confused with Daybreak Red, OpenAI's approved-access program for specialized cybersecurity models such as GPT-5.6 Cyber and GPT-6 Cyber. Daybreak Blue serves common defensive work with the mainline models; Red is restricted to authorized, more sensitive activities such as advanced vulnerability research and controlled exploit validation. Access to advanced capabilities remains limited to approved defenders, with safeguards matched to the activity.
Geekiki has already covered Daybreak Red's arrival in AWS GovCloud and the entry of Pax, the Brazilian "police AI", into Daybreak Blue: OpenAI takes Daybreak Red with GPT-5.6 Cyber to AWS GovCloud and Pax, the Brazilian "police AI", joins OpenAI's Daybreak Blue.
Image credit: official OpenAI illustration about code security review, published on the company's developers blog.
Sources
- OpenAI Developers — Scaling cyber defenders with Daybreak
- OpenAI — Daybreak
- OpenAI — Daybreak for Frontline Defenders: $1B to protect essential services
- The Register — OpenAI launches security products, tells orgs they're doomed if they don't use them (November 12, 2026)
- The Deep View — OpenAI sees a new enterprise frontier in cyber (November 12, 2026)
Transparency: this article was produced with the help of artificial intelligence, based on cross-checking public posts on X and internet sources, and verified against the original sources listed above. For full details, consult the original sources.
By Muse
