OpenAI's Daybreak cyber defense models now on AWS Bedrock

AWS and OpenAI announced that Daybreak Red and Daybreak Blue—specialized AI models for cybersecurity—are now available on Amazon Bedrock, AWS's managed service for running AI workloads. Daybreak Red provides access to GPT-5.6 Cyber, a model purpose-trained for cybersecurity. Daybreak Blue provides access to GPT-5.6 Sol, trained with safeguards calibrated for defensive cybersecurity work. Both are part of OpenAI's Daybreak initiative, which gives defenders governed access to frontier AI, including agentic tooling, application red teaming, and services that help move from findings to tested fixes.

The availability addresses a critical challenge: the window between when a vulnerability is disclosed and when it is exploited keeps shrinking. Frontier models can now reason across entire codebases, trace a vulnerability to its root cause, and propose a fix in minutes. However, cybersecurity is inherently dual-use—a request to reproduce a vulnerability or reverse-engineer an exploit chain looks identical regardless of intent. General-purpose models resolve that ambiguity by declining the request. Daybreak Red and Daybreak Blue resolve it through context: who is using the model, where the work occurs, and what safeguards govern access. Daybreak Blue is the right starting point for most security teams, supporting vulnerability discovery, detection engineering, and incident response. Daybreak Red is designed for advanced tasks like vulnerability research, exploit reproduction, and mitigation development, with a lower refusal threshold matched by stronger identity verification, monitoring, and access controls.

The models run on Amazon Bedrock's next-generation inference engine. Zero-operator access is enforced at the chip, so even AWS operators cannot access prompts and completions during inference. Everything is encrypted in transit and at rest with customer-managed AWS KMS keys. Access is governed by AWS Identity and Access Management policies, logged in AWS CloudTrail, and routed through VPC endpoints. Customers can set data perimeter policies at the organization level to prevent exfiltration. Inference data is not used for model training, and neither model requires opt-in to share data with OpenAI; classifier-flagged traffic for automated abuse detection is retained by AWS for up to 30 days and processed programmatically, though customers may request zero data retention through their AWS account team.

According to OpenAI, security researchers used GPT-5.6 Cyber through Daybreak Red to identify two previously unknown vulnerabilities in V8, the JavaScript engine used by Chrome. When chained together, they could enable memory corruption and a heap sandbox escape. The initial vulnerability was fixed and released as CVE-2026-15903, one of only four successful zero-day entries to V8 CTF in 2026. Both models are now available to eligible customers on Amazon Bedrock in US East (N. Virginia). Access requires enrollment in Trusted Access for Cyber from OpenAI; eligible customers must contact OpenAI or their AWS account team for approval. AWS Security Vice President John Sheehan noted that AWS security teams are already using both models to analyze source code, discover vulnerabilities, and conduct red-team research under the same infrastructure controls applied to other critical workloads.