infrastructure
Amazon Bedrock Introduces Private Key JWT Authentication for AgentCore Identity
Amazon Bedrock's AgentCore Identity now supports Private Key JWT client authentication, allowing agents to securely authenticate using signed assertions stored in AWS KMS. This update improves security and auditability for AI agent interactions with identity providers.
AS1 NewsSource: aws.amazon.com
Amazon Bedrock has announced the addition of Private Key JWT client authentication support within its AgentCore Identity service. This feature allows AI agents to authenticate to downstream identity providers using signed JSON Web Tokens (JWTs), with private keys securely stored and managed in AWS Key Management Service (KMS). Instead of relying on shared secrets, agents now sign assertions with KMS, which are then verified by the identity provider, enhancing security and reducing the risk of secret compromise.
The process involves creating an asymmetric KMS signing key, registering its public key with the identity provider, and configuring a credential provider in the AWS Management Console. When an agent requests a token, it constructs a JWT assertion, signs it with KMS, and sends it to the identity provider's token endpoint. The provider verifies the signature using the registered public key and issues an access token.
This method supports three grant flows: machine-to-machine (M2M), on-behalf-of (OBO), and user-delegated access, covering a broad range of AI and automation use cases. The approach offers a secure, auditable, and scalable way for AI agents to authenticate, especially in enterprise environments where security and compliance are critical.
The update also includes detailed steps for creating and managing KMS keys, registering public keys, and configuring OAuth clients within the AWS ecosystem. CloudTrail logs record each signing operation, providing transparency and auditability.
This enhancement is likely to impact organizations deploying AI agents at scale, improving security posture and compliance. It also demonstrates AWS's commitment to advancing secure AI infrastructure practices, which could influence industry standards and best practices for AI security.
This update improves security and auditability for AI agent authentication, supporting enterprise adoption of secure AI infrastructure.