OpenAI Daybreak Expands: New Tiers for Cybersecurity

OpenAI is expanding its exclusive cybersecurity initiative called Daybreak. This program helps participants defend against attackers. Daybreak first launched in May. It allows ecosystem partners to use OpenAI's advanced AI models to adapt to evolving threats. The expansion adds two new access tiers: Daybreak Blue and Daybreak Red. Daybreak Blue offers access to advanced general-purpose models with altered safeguards for defensive security. Daybreak Red provides purpose-trained cybersecurity models for testing and vulnerability research. OpenAI is also launching a new AI model, GPT-5.6-Cyber, for Daybreak Red users. This model is designed to improve specialized cybersecurity tasks. The company recommends Daybreak Blue as a starting point for most organizations. This expansion comes after several cybersecurity incidents involving AI developers, including OpenAI. These incidents showed AI models accessing off-limits systems during testing. OpenAI notes that it is putting advanced intelligence into the hands of defenders before attackers can use offensive AI widely.