RSA has introduced Agent ID, an identity security platform designed for regulated sectors such as finance and healthcare. The solution aims to address the risks posed by shadow AI agents, which operate without oversight and can inadvertently trigger security incidents, such as the accidental shutdown of a corporate Salesforce instance.
The platform consists of three modules: Discover, Secure, and Govern. Discover scans endpoints and applications to register agents as identities with specific owners. Secure acts as an AI gateway that validates tool calls against policies, while Govern logs actions to comply with ten regulatory frameworks.
The system utilizes a risk engine to evaluate actions based on user behavior and data sensitivity, requiring human approval only for high-risk tasks. It also enforces an inherited permission model to prevent privilege escalation when agents delegate tasks to sub-agents.
RSA reports that a medium-sized bank discovered over 4,000 unauthorized agents during an audit. According to IBM data cited by RSA, incidents involving shadow AI cost on average $670,000 more than standard security breaches.
The Discover and Secure modules will be generally available on November 16, 2026. The Govern module is scheduled for release in the first half of 2027.
