Entrust launches agentic AI accelerator to add cryptographic trust

Entrust this week launched an agentic AI accelerator designed to provide cryptographic identity, attestations and testing tools for autonomous agents. The product is available now through Entrust’s enterprise sales channels and partner network.

The offering packages software development kits, security modules and governance tooling for enterprises and developers building autonomous agents. Entrust brought its digital-identity and key-management capabilities into workflows for agent creation, deployment and monitoring to produce identity-bound actions and audit records.

Features include APIs to issue verifiable credentials to agents, hardware-backed key storage for signing agent actions, runtime integrity checks and centralized logging. The design provides tamper-resistant attestations of agent state, automated audit trails for agent actions and signed records that tie a specific key to a specific action.

Entrust positioned the accelerator for regulated and mission-critical environments where traceability and non-repudiation are required. The company described typical use cases as automated trade execution in financial services, control agents in energy and utilities, and agentic assistants that handle customer data or IT administration in enterprises.

The stack supports integration with existing identity systems, cloud providers and orchestration platforms so customers can add cryptographic identity and attestations without replacing core infrastructure. The package includes a developer sandbox and compliance templates to test agent behavior and produce evidence for audits.

Testing tools cover adversarial evaluation, scenario-based behavior testing and metrics for reliability and safety that can be exported for compliance reviews. Role-based controls let organizations limit an agent’s scope and require human approval for sensitive operations. Monitoring and alerting functions flag anomalous agent behavior and generate immutable logs and signed attestations intended for forensic review.

Entrust also outlined plans for third-party certification programs to validate vendor implementations against a common set of criteria. The company is known for supplying identity and security infrastructure to enterprises and governments, including certificate management and secure key services, and positioned the accelerator as an extension of those capabilities into agent lifecycle management.

Agentic AI refers to systems that can take autonomous actions, often chaining tasks and making decisions without continuous human supervision. Concerns about such systems include attribution of actions, unpredictable behavior, data protection risks and regulatory compliance. Cryptographic identity, attestations and immutable logging are technical methods that make agent actions verifiable and auditable.