On Monday, Sam Altman announced Daybreak, OpenAI's new cybersecurity initiative designed to "accelerate cyber defense and continuously secure software." The timing was conspicuous. It came one day before the Pentagon publicly confirmed it was deploying Anthropic's Claude Mythos through Project Glasswing to patch government networks.

The two most valuable AI companies in the world are now openly competing to become the default cybersecurity infrastructure for governments and corporations. The AI cybersecurity arms race, which was an anonymous Axios report just a month ago, is now a two-front war with public marketing campaigns.

What Daybreak Actually Does

Daybreak builds on Codex Security, a research preview OpenAI released in March. The system creates a "threat model" of a given organization by mapping its functions, trust relationships, and potential attack surfaces. It then scans the actual codebase for real-world exploits. In theory, it then patches them.

The approach is less dramatic than Anthropic's Glasswing rollout. Where Glasswing launched alongside Mythos Preview, a model so capable Anthropic refused to release it publicly, Daybreak's landing page has two buttons: "Request a vulnerability scan" and "Contact sales." The form is brief. The pitch is open. Altman said OpenAI would "like to start working with as many companies as possible now."

As developer Daniel Stenberg noted of Anthropic's approach, the Glasswing launch was "an amazingly successful marketing stunt." Daybreak reads like OpenAI's attempt to do the opposite: less mystery, more accessibility, fewer headlines about AI too dangerous to release.

Two Philosophies, One Market

The contrast between the two programs reveals a fundamental strategic split. Anthropic's playbook: build the most capable model possible, declare it too dangerous for general release, then partner exclusively with governments and major institutions under controlled access. The scarcity creates urgency. When the Pentagon, US banks, and the British government all rush to adopt Mythos, it validates Anthropic's position as the serious, safety-first lab.

OpenAI's playbook: build comparable capability, make it broadly available, and compete on reach rather than exclusivity. Daybreak is open to any company that fills out a form. The implicit argument is that cybersecurity tools should be democratized, not gatekept.

Both approaches have risks. Anthropic's exclusivity means many organizations that need cybersecurity AI cannot access it. OpenAI's openness means the same tools available to defenders could inform attackers. Neither company has solved the fundamental dual-use problem at the heart of AI cybersecurity.

The Stakes Keep Rising

The cybersecurity AI market is not theoretical. Banks are already scrambling to patch thousands of vulnerabilities that Mythos found in their systems. The Pentagon is deploying Mythos across government networks. Forbes reported this week that a startup found critical vulnerabilities that Mythos missed, suggesting even the most capable models have blind spots.

Meanwhile, Dario Amodei recently revealed that Anthropic "80x'd the business" rather than the 2-3x they expected, with cybersecurity partnerships driving a significant portion of that growth. The financial incentives for both labs to dominate this market are enormous.

The question is no longer whether AI will become the backbone of global cybersecurity. It already is. The question is whether the market will consolidate around one lab's approach or split between Anthropic's controlled-access model and OpenAI's open-access model. For CISOs and government CTOs making procurement decisions right now, the choice between Glasswing and Daybreak is not just a vendor decision. It is a bet on which philosophy of AI safety will win.