What happened
On Sept 24, 2026 Fortune (citing multiple sources) reported OpenAI is preparing to preview GPT-6 Cyber — its fourth cybersecurity-focused model of 2026, following GPT-5.4/5.5/5.6 Cyber — likely at DevDay on Sept 29, and to launch a new, as-yet-unnamed product that helps customers deploy the model more securely and automatically. Alpha testing is underway with Daybreak Red program customers.
Why it matters
A fourth frontier cyber model plus a dedicated secure-deployment product underscores OpenAI's push into enterprise cybersecurity (backed by a $1B subsidy commitment) and signals that AI-for-security models are becoming deployable, governed products rather than raw APIs — significant for the whole AI-security market.
Applicability
Security teams trialing AI for offensive/defensive work should watch the Sept 29 DevDay for the GPT-6 Cyber preview and the deployment-safety product; red teams in Daybreak Red can already alpha-test.