OPENAI EXPANDS DAYBREAK CYBERSECURITY SERVICE WITH TWO ACCESS TIERS

OpenAI has announced an expansion of Daybreak, its dedicated cybersecurity initiative, introducing two new access tiers named Daybreak Blue and Daybreak Red. The expansion includes a new security-focused AI model, GPT-5.6-Cyber, designed for authorised vulnerability research. OpenAI stated that access to these tools remains restricted due to dual-use risks and reduced safeguard operation. The company also confirmed that Daybreak Blue serves as the recommended starting point for most defenders, supporting vulnerability discovery, secure code review, malware analysis, incident response and patch validation.

Daybreak was initially launched in June 2026 and originally provided vetted cybersecurity companies with state-of-the-art models without standard content guardrails. The original initiative included GPT-5.5-Cyber, an agent called Codex Security capable of analysing codebases and identifying vulnerabilities, and Patch the Planet, a partnership with Trail of Bits focused on finding and fixing vulnerabilities in open-source software. The Daybreak Cyber Partner Program allows approved companies such as Cloudflare and Cisco to integrate OpenAI's capabilities into their own products. OpenAI created Daybreak because guardrails designed to prevent attackers from using AI to build malware also slowed cybersecurity defenders in their work.

The expansion comes amid increased use of AI by criminal actors to automate attacks, create phishing emails and generate malicious code. Anthropic released a similarly cyber-focused model called Mythos earlier this year. OpenAI's tiered approach aims to provide the cybersecurity community with access to advanced defensive tools whilst managing the risks associated with providing unrestricted AI capabilities.