OpenAI put up product pages on 12 November, the day of its Cyber at the Frontier keynote, for Daybreak Red Team, which its page calls “AI penetration testing for web applications” and lists as early access, and Daybreak Defense Factory, which “Put AI agents to work investigating vulnerabilities and preparing patches for review”. A third page sets up a Daybreak plugin for Codex.
The launch comes 69 days after OpenAI committed $1 billion in subsidized Daybreak access, which we covered on 4 September. Neither product page states a price; the Red Team page asks buyers to “Talk to our team about early access to Daybreak Red Team”.
Business Pill · Least Privilege
A short explainer on giving an AI agent only the access its task needs. It teaches the general idea only and does not discuss OpenAI’s Daybreak products.
The key insight: As we read the pages, OpenAI has turned Daybreak from an access program into products that do the security work itself: Red Team agents that test web apps and check whether fixes hold, in early access, and Defense Factory agents that prepare patches for engineers to review, with no price on either page.
The Red Team page says it will “Put a coordinated team of agents to work testing your web applications, validating vulnerabilities, and checking if patches prevent the exploit”. “Agents use source code and app context to investigate attack paths and confirm vulnerabilities”, it says.
It also tests findings that come from elsewhere: “code scanners, bug bounty submissions, and disclosure reports”. After a fix, it can “Test whether a patch stops the exploit” and repeat the assessment as the apps change.
On control, the page says buyers can “Enforce assessment scope with isolated environments and Guardian review” and choose “which apps and endpoints agents can reach”.
What OpenAI’s Daybreak page, the new Red Team and Defense Factory pages and its access guide list on 12 November, by our count. Neither new product page states a price.
What Defense Factory Does
The Defense Factory page says “Agents trace vulnerabilities through your code and develop patches. Engineers review and test proposed patches before merge.” Findings from a company’s existing security tools and from Daybreak Red Team feed into it, and agents investigate each issue and develop a fix.
It is built to run continuously: “Scan new commits or schedule recurring checks as your code changes”. Engineers can work on a patch with Codex “while agents continue working in the background”, the page says.
The name is not new. In its Frontline Defenders post, OpenAI said it had published its approach to building a Defense Factory, described as “a continuous, agent-first operation that builds on existing security and engineering tools to find and validate vulnerabilities and prepare tested fixes for review”. The 12 November page sells that approach as a product.
The Codex Plugin and the Access Gate
The plugin page offers “Desktop Codex for a guided flow, or Codex CLI for a one-command scan from a folder of code”. The Daybreak page lists four cybersecurity products: the Daybreak plugin, Defense Factory, Red Team, and the Codex Security CLI and SDK.
Daybreak itself has two access tiers, Blue for defensive work and Red for authorized offensive testing. OpenAI’s access guide says Daybreak Red “Requires separate Red approval, stronger verification, and access controls”, and that “OpenAI reviews requests before enabling access. Approval is not automatic.” Daybreak does not allow “resale, proxying, embedding, or downstream access for third-party customers or external users”, the guide says.
The Red tier is also where OpenAI’s cyber models sit. On 7 November we reported that OpenAI’s API page lists GPT-6-Cyber at $20 per million input tokens and $100 per million output tokens, twice GPT-6 Astra’s $10 and $50 by our arithmetic, with Daybreak Red approval required.
OpenAI’s API list prices per million tokens for GPT-6 Astra and GPT-6-Cyber, which needs Daybreak Red access and approval, as we reported on 7 November. The 2x ratio is our arithmetic.
The Installed Base and the Partners
OpenAI’s Frontline Defenders post said “Thousands of defenders across 2,000 approved organizations and workspaces already use Daybreak”, and that partners were announcing “more than 35 partner products and partner-operated services” built on its cyber models. The $1 billion is targeted to be consumed over six months, the post says.
Two days before the keynote, an OpenAI customer story said Cogent “builds Cyber Defense Factories for enterprises” on OpenAI models. Cogent reports that a Fortune 1000 automotive customer “reduced its mean time to remediate critical vulnerabilities by 75%”. OpenAI’s own product now carries the Defense Factory name.
The Structural Read
As we read the Defense Factory page, OpenAI is now selling under its own name an approach it published alongside its Frontline Defenders commitment, a name its customer Cogent also uses for what it builds on OpenAI models.
As we read the access guide, the gate is part of the offer as much as a safeguard: Red tier use needs separate approval, and the guide bars resale, proxying or embedding for third parties.
OpenAI, Daybreak Red Team product page, 12 November 2026
“Put a coordinated team of agents to work testing your web applications, validating vulnerabilities, and checking if patches prevent the exploit.”
Three Implications
FOR SECURITY TEAMS Red Team is in early access through OpenAI’s team, and Defense Factory takes findings from existing tools and Red Team and prepares patches that engineers review before merge, the pages say.
FOR OPENAI’S PARTNERSOpenAI said in its Frontline Defenders post that more than 35 partner products and services use its cyber models; its own Defense Factory product now shares a name with what Cogent builds.
FOR OUR COVERAGE We will add prices, availability dates or customer figures if OpenAI publishes them.
The framework puts it this way: “AGaaS charges for the execution of outcomes.”
As we read it, agents that run a penetration test or prepare a patch sell executed work rather than access. The framework predates the launch and does not discuss it, and OpenAI has not published how it will charge.
What Is Not Established
The pages give no price, no general-availability date and no customer count for either product. The keynote video on OpenAI’s YouTube channel has no captions available to us, so we quote nothing from it. A claim on X comparing OpenAI’s cyber model with a rival’s does not appear in any OpenAI text we read, and we do not use it.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
OpenAI turned Daybreak from an access program into a product line on 12 November: Red Team agents that test web apps and check fixes, in early access, and Defense Factory agents that prepare patches for engineers to review. Neither page states a price, 69 days after the $1 billion pledge.
A note on sourcing. We describe the products from OpenAI’s Daybreak Red Team, Defense Factory and plugin pages, updated on 12 November 2026, its Daybreak program page and its Help Center access guide. Installed-base and partner figures come from OpenAI’s Frontline Defenders post; the Cogent figure is Cogent’s, as published by OpenAI. GPT-6-Cyber prices come from our 7 November reporting of OpenAI’s API pages. The keynote video had no captions available, so nothing is quoted from it. Counts and the 69-day interval are ours.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
