The announcement arrives just weeks after Anthropic unveiled Project Glasswing, a restricted-access cybersecurity program tied to its Claude Mythos Preview model. Anthropic described that system as its “most cyber-capable model” and said the model would not be made broadly available because of concerns tied to its advanced capabilities. Instead, the company limited access to select partners working on defensive cybersecurity efforts.
OpenAI’s rollout takes a noticeably different approach. Rather than positioning Daybreak as a tightly controlled initiative available only to a small circle of organizations, the company is openly soliciting participation through its website. The launch page includes options to “Request a vulnerability scan” and “Contact sales,” and Altman said on X that OpenAI would “like to start working with as many companies as possible now.”
The project builds on Codex Security, a system OpenAI introduced in research preview form earlier this year. According to the company, Daybreak uses Codex Security to generate a “threat model” for a system by mapping how software functions, identifying trusted actors within the environment, and analyzing potential attack surfaces. From there, the system reviews codebases for vulnerabilities and possible real-world exploits.
OpenAI also says Daybreak can patch discovered vulnerabilities, though the company provided limited technical detail about how those fixes are implemented in practice.
The launch highlights how leading AI companies are increasingly competing beyond chatbot performance and model benchmarks, moving into infrastructure and enterprise security tools as businesses look for ways to safely deploy AI systems at scale. Security concerns surrounding prompt injection, model misuse, and software vulnerabilities have become more prominent as AI adoption spreads across corporate environments.
So far, neither OpenAI nor Anthropic has disclosed extensive technical information about the underlying capabilities of their cybersecurity systems or how broadly the tools will be deployed. But the near-simultaneous launches suggest major AI firms see defensive security products as an emerging area of competition tied closely to enterprise adoption.
This analysis is based on reporting from Gizmodo.
Images courtesy of OpenAI.
This article was generated with AI assistance and reviewed for accuracy and quality.
