OpenAI's Daybreak adds 2 tools to fix flaws

OpenAI launched two security tools on the 12th of November 2026, the company said. Both use AI agents to find and fix weak spots in company software.

OpenAI President Greg Brockman said companies will be secure only to the extent of the AI model they use to defend themselves, The Register reported.

Daybreak Red Team tests web apps

Daybreak Red Team sends a team of AI agents to attack a company's web apps and look for flaws. It also checks bug reports and retests fixes, OpenAI's page says.

The tool is in early access, The Register and RuntimeWire reported.

Defense Factory writes the fixes

Daybreak Defense Factory hunts for flaws in a company's code and drafts patches, which are small code fixes. Engineers review and test each patch before it goes in, OpenAI said.

Its agents run on GPT-6 Cyber, OpenAI's latest security model, The Register reported.

A free tool gets an update

OpenAI also offers Codex Security, a free tool with open code that scans software for flaws and suggests fixes. Version 0.10.0 came out on the 12th of November, its GitHub page shows.

The GitHub page dates from July 2026, so this is an update.

Below we break down the cost

GPT-6 Cyber costs 20 dollars per million input tokens and 100 dollars per million output tokens, VentureBeat reported. A million of each costs 120 dollars, by our arithmetic.

That is 10 times the 12 dollars for GPT-6 Sol.

What is confirmed and what is not

OpenAI's pages give no prices for either Daybreak tool, and no date for wider access.

Key facts

OpenAI launched Daybreak Red Team and Daybreak Defense Factory on the 12th of November 2026.

Red Team is in early access.

Engineers approve every fix before it goes in.

Codex Security is free, with open code.