New to this? Read it in simple words
- OpenAI announced two security products with AI agents on November 12.
- Red Team attacks a company’s own web apps to find real flaws.
- Defense Factory writes patches, and engineers must review them before use.
- OpenAI has not given prices, and no outside group has tested them yet.
- Red team
- People, or here AI agents, who attack an organisation’s own systems with permission to find weak spots.
- Patch
- A change to software code that fixes a flaw.
- Pull request
- A proposed code change that engineers review before they add it to the main code.
A red team attacks a company’s own systems, with permission, to find weak spots before criminals do. OpenAI now offers AI agents for that job, and more agents to fix what they find.
Red Team: agents that attack your app
Daybreak Red Team tests web applications. OpenAI says its agents “use source code and app context to investigate attack paths and confirm vulnerabilities”.
A fleet of agents does the work, The Register reports. Red Team can also check flaws reported by code scanners, bug bounty hunters and outside researchers, and test whether a patch stops the attack.
OpenAI says the agents work in isolated environments, and that companies control which apps and addresses they may reach.
How the two products hand work to each other, from OpenAI’s pages.
Defense Factory: agents that prepare fixes
Defense Factory takes flaws found by Red Team or by other security tools, and its agents trace each one through the code and write a patch. SecurityWeek says it can bring in findings from Wiz, Snyk and Endor Labs.
It can scan connected GitHub code, check new changes, or run on a schedule. Engineers can tune a patch with Codex, OpenAI’s coding agent, and open a draft change for review. Patches are not deployed by themselves.
“This way engineers get a merge-ready pull request with the patch and a test that shows that it works,” said Nancy Wang, who leads OpenAI’s applied cyber work, as The Register reports.
Who can use it
Red Team is in early access for selected customers, with a waiting list. Defense Factory is a preview plugin for organisations with Daybreak access; The Deep View places it in Codex and SecurityWeek in ChatGPT workspaces.
OpenAI has not published prices. The Register reports that Defense Factory runs on GPT-6 Cyber, the security model OpenAI released to approved customers on November 5.
What OpenAI says, and what is missing
OpenAI president Greg Brockman told the event that companies “will be secure only to the extent of the frontier model that they use to defend themselves”. He also said companies must “reorient” their security work around what these models can do.
Johannes Landgraf, OpenAI’s head of cyber, said attackers “are able to move faster” and that defenders must too. These are the company’s views. No outside test of either product has been published, and The Deep View notes that security could also be a new source of income for OpenAI.
Sources
Every fact in this story comes from the sources below. Open them to check our work.
- 1
- 2
- 3Research · November 12, 2026OpenAI launches security products, tells orgs they’re doomed if they don’t use them The Register
- 4
- 5Research · November 13, 2026OpenAI Unveils Daybreak Red Team, Defense Factory Security Tools SecurityWeek
What the agents do comes from OpenAI’s product pages and its staff. How well they find and fix flaws is OpenAI’s claim, and no outside test has been published.
OpenAI’s new Daybreak agents test apps and draft fixes. Silicon AI News, November 14, 2026. https://siliconainews.com/stories/openai-daybreak-red-team-defense-factory
OpenAI launched AI agents that find real app flaws and draft reviewed fixes, though no outside test has checked them yet. How we rate the mood
The mood shows how the news frames the event for the tech and AI world: as an opportunity (6 and above) or as a loss (below 5). It is not our verdict on who is right.