Which AWS region and models are available now?

Daybreak Red (GPT-5.6 Cyber) and Daybreak Blue (GPT-5.6 Sol) are available in US East (N. Virginia). Daybreak Blue is the starting point for most teams; Daybreak Red is for advanced tasks like vulnerability research and exploit reproduction.

How is my sensitive code and vulnerability data protected?

Both models run on Amazon Bedrock's next-generation inference engine with zero-operator access enforced at the chip level. Data is encrypted in transit and at rest using customer-managed AWS KMS keys, logged in AWS CloudTrail, and routed through VPC endpoints. Your inference data is not used for model training unless you opt in.

What example success has OpenAI shared?

Security researchers using GPT-5.6 Cyber through Daybreak Red identified 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.

Large Language ModelsAI Safety & AlignmentAI Business & IndustryAmazon AI BlogAug 12, 2026Aug 12, 2026, 10:00 JST

OpenAI's Daybreak cyber defense models now on AWS Bedrock

Send on LINE

3 Key Points

What happened
OpenAI's Daybreak Red (GPT-5.6 Cyber) and Daybreak Blue (GPT-5.6 Sol) are now available to eligible customers on Amazon Bedrock in US East (N. Virginia). Both models are purpose-built for cybersecurity work—Daybreak Red for advanced tasks like vulnerability research and exploit reproduction, Daybreak Blue for vulnerability discovery and incident response.
Why it matters
Security teams can now analyze source code, identify vulnerabilities, and conduct red-team research while keeping sensitive code and vulnerability data encrypted and isolated within AWS infrastructure they control. The models can reason across entire codebases and propose fixes in minutes, helping defenders close the shrinking window between vulnerability disclosure and exploitation. According to OpenAI, security researchers using GPT-5.6 Cyber identified two previously unknown vulnerabilities in V8 (Chrome's JavaScript engine) that, when chained, could enable memory corruption and a heap sandbox escape—one of only four successful zero-day entries to V8 CTF in 2026.
What to watch
Access requires enrollment in Trusted Access for Cyber from OpenAI; eligible customers must contact OpenAI or their AWS account team for approval. Zero-operator access is enforced at the chip level, inference data is encrypted in transit and at rest using customer-managed AWS KMS keys, and data is not used for model training unless customers opt in. Customers may request zero data retention through their AWS account team.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Get news like this every morning, free

Context & Analysis

The cybersecurity landscape has shifted in recent years. Frontier AI models can now reason across entire codebases, trace vulnerabilities to their root cause, and propose fixes within minutes—capabilities that belong in defenders' hands but also in adversaries'. The body emphasizes that the window between vulnerability disclosure and exploitation is shrinking, placing immense pressure on security teams to validate findings and ship patches quickly across sprawling, unfamiliar code bases. The challenge is no longer surfacing potential issues but confirming which findings are real and acting before the window closes.

AWS and OpenAI are addressing this by bringing Daybreak Red and Daybreak Blue to Amazon Bedrock, where they operate under the same infrastructure, controls, and governance customers already trust for other AWS workloads. The dual-use nature of cybersecurity work—where the same request to reproduce a vulnerability or reverse-engineer an exploit chain could come from either defender or attacker—demands context-aware safeguards. Daybreak Blue starts with a higher refusal threshold and is designed for broader defensive tasks; Daybreak Red lowers that threshold for advanced researchers but enforces stronger identity verification, monitoring, and access controls. The body demonstrates this approach works: researchers using GPT-5.6 Cyber identified two previously unknown V8 vulnerabilities that became one of only four successful zero-day entries to V8 CTF in 2026.

FAQ

Which AWS region and models are available now?: Daybreak Red (GPT-5.6 Cyber) and Daybreak Blue (GPT-5.6 Sol) are available in US East (N. Virginia). Daybreak Blue is the starting point for most teams; Daybreak Red is for advanced tasks like vulnerability research and exploit reproduction.
How is my sensitive code and vulnerability data protected?: Both models run on Amazon Bedrock's next-generation inference engine with zero-operator access enforced at the chip level. Data is encrypted in transit and at rest using customer-managed AWS KMS keys, logged in AWS CloudTrail, and routed through VPC endpoints. Your inference data is not used for model training unless you opt in.
What example success has OpenAI shared?: Security researchers using GPT-5.6 Cyber through Daybreak Red identified 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.

Amazon AI BlogRead Original Article

Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.