AWS and OpenAI have expanded their collaboration to enhance cyber defense capabilities by making Daybreak Red and Daybreak Blue available on Amazon Bedrock. These models are designed to help security teams identify and address vulnerabilities more efficiently. The release comes as the window between vulnerability disclosure and exploitation continues to shrink, placing greater pressure on defenders to act quickly and effectively. According to AWS, the partnership aims to give defenders an advantage by providing access to frontier AI models that can analyze code, discover vulnerabilities, and support red-team research. 'AWS and OpenAI share a belief that defenders should have the advantage,' said John Sheehan, Vice President, AWS Security. 'This partnership brings Daybreak Red and Daybreak Blue from OpenAI to Amazon Bedrock. AWS security teams are using both models today to analyze source code, discover vulnerabilities, and conduct red-team research.'

Daybreak Red provides access to GPT-5.6 Cyber, a purpose-trained cybersecurity model, while Daybreak Blue offers GPT-5.6 Sol with safeguards tailored for defensive cybersecurity work. Both models are part of the Daybreak initiative, which gives defenders governed access to advanced AI tools, including agentic tooling, application red teaming, and services that help move from findings to tested fixes. 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, which when chained together 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.

The models run on Amazon Bedrock's next-generation inference engine, which is built for high performance, security, and reliability. Zero-operator access (ZOA) is enforced at the chip level, ensuring even AWS operators cannot access prompts or completions during inference. All data is encrypted in transit and at rest using customer-managed AWS Key Management Service (AWS KMS) keys. Access is governed by AWS Identity and Access Management (IAM) policies, logged in AWS CloudTrail, and routed through virtual private cloud (VPC) endpoints. Customers can set data perimeter policies at the organization level to prevent exfiltration across account and network boundaries. Inference data is not used for model training, and neither model requires customers to opt into sharing data with OpenAI.

Source: awsml