OpenAI Expands Daybreak Cyber Defense With New GPT-5.6-Cyber Model

OpenAI announced on Monday, August 10, an expansion of Daybreak, its cyber defense service launched earlier this year, adding a new purpose-trained cybersecurity model called GPT-5.6-Cyber as AI-led attacks multiply, TechCrunch reported.

Two Tiers: Blue and Red

Daybreak now consists of two tiers. Blue, described by OpenAI as the 'recommended starting point for most defenders,' bundles incident response, malware analysis and patch validation for enterprises. Red offers a broader toolkit for security testing and vulnerability research, including purpose-trained cybersecurity models, and is the only tier that includes GPT-5.6-Cyber, a specialized model built on GPT-5.6 Sol.

Trusted Partners Only

GPT-5.6-Cyber is initially available only to approved customers, reportedly including Accenture, IBM, CrowdStrike and Cloudflare, per BleepingComputer. The launch follows a series of high-profile incidents in which AI agents behaved like bad actors — from a Hugging Face breach to agents creating fake profiles to socially engineer an intrusion, as reported by TechCrunch and the BBC.

'The cybersecurity world is rapidly changing — threat actors will increasingly use AI to conduct cyberattacks at unprecedented speed and scale, including in fully autonomous ways,' OpenAI said in a blog post. 'As these capabilities spread, defenders have a narrowing window to prepare.'