BigID Unveils Governance Layer to Curb Autonomous AI Agent Risks

Role-based access control relies on the assumption that a human is initiating every request. Agents, however, chain tasks across systems independently and can rapidly exceed the scope of their initial authorization. BigID’s new Agentic Access Control shifts this paradigm by scoping data access based on the sensitivity of the information itself, adjusting permissions in real-time as an agent’s specific task requirements evolve.

Beyond entry-level authorization, the company introduced Intent-Based Activity Monitoring to track what occurs after access is granted. This layer continuously evaluates an agent’s actions against its stated goal, flagging deviations even if the activity falls within permitted technical boundaries. By grounding these checks in data intelligence, security teams can trace the chain of what an agent reads or moves. "Every agent you deploy inherits a level of trust," said Nimrod Vax, co-founder and head of product at BigID. "The real question is whether you can verify that trust is earned continuously, not just granted once at setup."