OpenAI expanded its Daybreak cybersecurity program on August 10, 2026, introducing two access tiers: Daybreak Blue and Daybreak Red, alongside a new AI model named GPT-5.6-Cyber [1, 2, 3, 4].

Daybreak Blue grants access to advanced general-purpose AI models with modified safeguards that enable defensive security operations. OpenAI recommends this tier as the starting point for most organizations aiming to strengthen their cybersecurity posture [1, 2, 3, 4].

Daybreak Red offers access to purpose-trained AI models designed specifically for security testing, vulnerability research, and exploit validation. This tier also provides exclusive use of GPT-5.6-Cyber, a model built on the GPT-5.6 Sol architecture. GPT-5.6-Cyber is tailored to improve performance on highly specialized cybersecurity tasks and reduce refusals on sensitive queries [1, 2, 3, 4].

OpenAI is collaborating with trusted partners including Accenture, IBM, CrowdStrike, Cloudflare, Cisco, and Sophos to deliver these enhanced capabilities under the expanded Daybreak program [2, 3].

The expansion follows OpenAI’s decision to pause some internal activities related to a forthcoming model called Astra. Concerns about Astra’s agentic coding abilities and potential to develop zero-day exploits led to the halt, reflecting heightened caution amid recent cybersecurity incidents where AI models accessed unauthorized systems [1, 3, 4].

OpenAI said, "As the threat landscape evolves, we're putting frontier intelligence in the hands of trusted defenders before attackers can deploy offensive AI at scale" [1]. The company also emphasized its commitment to responsible AI deployment, stating, "We're committed to working alongside governments, security agencies and civil society to ensure responsible deployment of Astra and future frontier models that broadly benefit humanity" [4].

OpenAI launched the initial Daybreak initiative in May 2026 to provide safer AI tools for cybersecurity tasks. The August 10 expansion adds new layers of access and specialized models.

The next key milestone will be further development and deployment updates related to Astra, which remains paused amid ongoing evaluations of its security implications [1, 3, 4].