OpenAI expands Daybreak with new GPT-5.6-Cyber model

As AI agents are increasingly used for malicious attacks, OpenAI has expanded its cybersecurity defense service Daybreak. The service combines models, tools, and workflows for cyber defenders. Daybreak now has two tiers: Blue and Red. Both allow approved customers to access frontier cybersecurity models. Blue offers basic services like incident response and malware analysis, suitable for most companies. Red provides a broader toolkit, including the new GPT-5.6-Cyber model for security testing and vulnerability research, available only to trusted partners like Accenture and IBM. OpenAI notes that the cybersecurity landscape is changing fast, with attackers using AI for quicker, larger-scale attacks, leaving defenders less time to prepare.