OpenAI expands cyber defense platform with new specialized AI model

As reports of AI-driven breaches accumulate—from compromised datasets at Hugging Face to attacks on gym websites and coordinated social engineering campaigns—major AI labs are responding with new defensive capabilities. OpenAI this week rolled out an expanded version of its Daybreak platform, a cyber defense suite that bundles model access, tooling, and workflows for security teams.

Two tiers for different protection needs#

The revamped offering comes in two tiers. The basic tier, called Blue, targets enterprises looking for standard defensive capabilities like incident response, malware identification, and patch testing. OpenAI positions this as the starting point for most defenders.

The premium tier, Red, unlocks deeper defenses. It provides access to purpose-trained cybersecurity models built specifically for security testing and vulnerability discovery. Only Red tier customers gain access to OpenAI's latest addition: GPT-5.6-Cyber, a specialized model derived from GPT-5.6 Sol and tuned for advanced cybersecurity tasks.

Limited early access#

For now, the new cyber model is available only to approved partners—a roster that reportedly includes Accenture, IBM, Crowdstrike, and Cloudflare. The restriction mirrors how other frontier AI models have been managed, following pressure from the Trump administration over safety and control concerns. OpenAI previously imposed strict guardrails on access to its most advanced models.

The competitive and marketing angle#

The move arrives months after Anthropic launched its own cyber-focused model, Mythos. Observers note that while the threat landscape from autonomous AI agents is genuinely widening—prompting urgent defense—AI labs are simultaneously seizing a business opportunity. Companies view the labs themselves as credible providers of protection because those same labs understand the risks from firsthand experience. The messaging around Daybreak's expansion underscores this framing: "The cybersecurity world is rapidly changing—threat actors will increasingly use AI to conduct cyberattacks at unprecedented speed and scale," OpenAI wrote, emphasizing the narrowing window defenders have to prepare.