OpenAI launches new cyber model as AI-led attacks multiply, expanding Daybreak program

OpenAI is expanding its AI cybersecurity defense program Daybreak and rolling out a new cyber-trained AI model with it, according to TechCrunch.

The move comes as AI-driven cyberattacks are multiplying rapidly, leaving security teams in need of stronger tools to detect and respond to automated threats. OpenAI aims to give defenders more efficient identification and response capabilities through a dedicated cyber model.

Daybreak is OpenAI's program focused on cybersecurity defense, and this expansion means more security researchers and enterprises can access its capabilities. The new model is trained specifically for cybersecurity, distinguishing it from general-purpose models.

The report does not disclose the new model's name, parameter count or availability, but describes it as a significant investment on the defensive side. Its real-world effectiveness will need to be tested in actual attack and defense scenarios.

For the security industry, the significance is that leading AI companies are now treating cybersecurity as a dedicated model track rather than an add-on to general models, reflecting how AI offense-defense dynamics are becoming a core battleground for model vendors.

What to watch next: whether the model will be opened to security vendors via API, which partners join Daybreak, and how OpenAI balances the model's potential use on both offense and defense. Those details will determine its real-world impact.