From Astra to Cyber: How OpenAI’s Safety Crisis Became a Product Strategy

OpenAI shipped four distinct cybersecurity models — GPT-5.4, 5.5, 5.6, and the recently previewed GPT-6 Cyber — within a single 12-month window. This rapid cadence forced a fundamental shift in strategy: the primary challenge for frontier labs is no longer mo