OpenAI CEO Sam Altman arrives at the White House in Washington, DC, USA for a conference, July 30, 2026.
Nathan Howard | Reuters
OpenAI The company announced Monday that it is expanding its proprietary cybersecurity initiative, Daybreak, to give participants access to the “appropriate capabilities” they need to protect themselves from attackers.
The company first introduced Daybreak in May, shortly after its main rival Anthropic attracted Wall Street and the U.S. government by launching its own cybersecurity coalition called Project Glasswing. OpenAI has positioned Daybreak as a way for ecosystem partners to use their cutting-edge artificial intelligence models to adapt to the rapidly changing threat landscape.
OpenAI announced Monday that it is expanding the program to include two different access tiers: Daybreak Blue and Daybreak Red. This extension supports OpenAI, human and Meta — revealed in recent weeks. In both cases, the AI models accessed systems that were supposed to be off-limits as part of cybersecurity testing, prompting industry researchers and government officials to call for stronger protections.
“As the threat landscape evolves, we are putting cutting-edge intelligence in the hands of trusted defenders before attackers deploy offensive AI at scale,” OpenAI said in a post on X on Monday.
Daybreak Blue gives users unique access to OpenAI’s advanced general-purpose models as protections are modified to enable defensive security work, the company said in a blog post. Daybreak Red participants can take it a step further by leveraging OpenAI’s “purpose-trained cybersecurity models” for security testing, vulnerability research, and exploit validation.
OpenAI also releases a new AI model, GPT‑5.6‑Cyber, available to Daybreak Red users. The company says this model is built on its most powerful publicly available product, GPT-5.6 Sol, but is designed to improve functionality and reduce denials for certain specialized cybersecurity tasks.
The company said it recommends Daybreak Blue as a starting point for most organizations.
OpenAI has repeatedly warned in recent months about the evolving cyber capabilities of its AI models. The company announced last week that it would suspend some internal activities related to its upcoming model, called the Astra, which demonstrated “significant advances in agent coding and cybersecurity” during testing.
It said it is evaluating these capabilities and working to introduce more robust safety measures and security controls.
“We are committed to working with governments, safety agencies and civil society to ensure that the cutting-edge capabilities of models like Astra, and those that follow, are responsibly and widely deployed for the benefit of all humanity,” OpenAI said in a statement.
Spotlight: OpenAI, Anthropic agents participate in new ‘unsanctioned’ AI operations
