Amazon Bedrock Adds OpenAI's Daybreak Cybersecurity Models for Approved Customers

At a glance

Amazon Bedrock, AWS's AI model service, now offers two cybersecurity-focused OpenAI models. Only customers approved for access can use them.

Key points

1. 1

   Daybreak Blue is the main model for defensive work such as vulnerability discovery, intrusion detection, and incident response.

2. 2

   Daybreak Red is for authorized advanced security work, such as vulnerability research and developing fixes.

3. 3

   Daybreak Red has a lower refusal threshold but adds identity checks, monitoring, and access controls.

4. 4

   Your input data is not used for model training, and you do not need to separately agree to share data with OpenAI.

Why it matters

Security teams now have a cloud-based AI tool for defending systems and analyzing vulnerabilities.