Grouped story
OpenAI Unveils New Cyber Defense Model Amid Rising AI Threats
OpenAI has expanded its Daybreak cyber defense service, introducing two tiers, Blue and Red, to combat the increasing threat of AI-led cyber attacks. The new GPT-5.6-Cyber model, designed for offensive security tasks, is available exclusively to trusted partners like IBM and Cloudflare, and has already identified multiple vulnerabilities in popular software.
Key points
Daybreak's Blue tier provides basic cyber services such as incident response and malware analysis, making it suitable for most enterprises.
Identity verification and security measures, including mandatory hardware keys, will be required for both Daybreak tiers by September 1, 2026.
GPT-5.6-Cyber has identified at least five vulnerabilities in a popular mobile operating system and two unknown vulnerabilities in Chrome's V8 engine, designated as CVE-2026-15903.