Cryptographic agent identity (PKI) for all agents
Control objective
Cryptographic agent identity gives every agent a verifiable, machine-checkable identity using public-key infrastructure (PKI), so that when agents talk to each other or to bank systems, each one can prove it is who it claims to be. Without strong identity, a malicious or spoofed agent could impersonate a trusted one and issue instructions - a serious threat in agent-to-agent (A2A) workflows that move money or release data. To implement, issue each agent an X.509 certificate from a managed certificate authority, require cryptographic authentication on every A2A message (not just at session start), rotate credentials on a tight schedule, and log identity issuance, message authentication and rotation events as evidence; integrate this with the bank's existing PKI and secrets management. No statistical metric applies; verification is cryptographic. The threshold has three parts: 100% of agents hold a valid X.509 identity, per-message authentication is enforced on all A2A traffic, and credentials are rotated within 24 hours. Any agent without a valid certificate, any unauthenticated A2A message, or credentials older than 24 hours is a breach that must block the agent or message and trigger reissuance, since it opens the door to impersonation.
Issue X.509 certificates to all agent instances. Require mutual TLS for all A2A communication. Rotate agent credentials every 24 hours or per session. Maintain centralised Certificate Authority for agent identity. Implement per-message authentication — do not trust identity once, verify per message.