AWS and Keyfactor released a solution using X.509 public key infrastructure (PKI) to secure AI agent identity on AWS, saying it addresses zero-trust requirements for multi-agent systems. It is the companies' first joint effort in AI security since their partnership announcement in 2023.
The solution reported four complementary security mechanisms, including signed request envelopes, user-signed delegation assertions, RFC 8693 audience-scoped tokens, and real-time OCSP revocation checking. That compares with traditional approaches such as API key sharing or OAuth token passing, which fall short of enterprise zero-trust requirements.
The solution is built on AWS services like Amazon Bedrock and Keyfactor EJBCA, targeting secure AI agent communication with external tools and data sources. Availability begins with a reference architecture, initially for developers and enterprise users.
"The architecture implements three authentication layers, each addressing a different trust boundary in the agent communication chain," said Keyfactor's EJBCA Enterprise product lead. The mechanisms combine across layers as shown in the following table.
The announcement follows AWS's push for secure AI deployment, as the source itself frames the significance of securing AI agent identity to prevent impersonation and unauthorized tool invocation.
AWS did not say how the solution will scale to larger agent networks, and the open question or limitation the source raises is whether the approach can be extended to non-MCP-compatible tools. The joint reference architecture, developed collaboratively by the AWS and Keyfactor engineering teams, demonstrates how to integrate these technologies.
Source: awsml
