#USA #Miami #Data Security #AI tools #Varonis

Varonis Introduces Agent Intent-Based Access Control

On August 3, 2026, Varonis Systems, Inc. (Nasdaq: VRNS) announced the launch of Agent Intent-Based Access Control (Agent IBAC) as a new feature of Varonis Atlas. This innovative capability ensures that businesses can securely connect AI agents to their enterprise data while implementing safeguards against risky or policy-violating behaviors. With incidents of AI agents inadvertently leaking confidential data or even deleting entire production databases, the need for reliable security measures has never been clearer.

One of the primary challenges is that AI agents require extensive access to data and tools to be effective, and this access creates potential risks. Traditional role-based access control systems were not designed to assess actions taken by non-human identities (NHIs). Therefore, Varonis has moved beyond asking, "Can a user access this data?" to questioning whether, in a given situation, an agent should be allowed to perform actions on that data.

Ron Bennatan, VP of AI and Data Security Strategy at Varonis, emphasized the importance of this shift in perspective. "We’ve developed Agent IBAC to provide organizations with the assurance that agents operate as intended and do not jeopardize the business."

How Agent IBAC Works

Agent IBAC verifies whether an agent's reasoning and tool utilization align with the assigned instructions, issuing warnings or blocking operations that do not match predefined criteria. If an agent crosses a line, Atlas can quarantine the underlying ID, blocking all subsequent actions for a defined period.

Key Features

• - Intent Drift Detection: This feature identifies discrepancies between the instructions received by an agent and its actual reasoning processes, tool usage, and data access. Users can set detection sensitivity to one of three levels: relaxed, standard, or strict.

• - Continuous Session Evaluation: Agent IBAC diligently examines all prompts, responses, and tool calls within a session to detect gradually accumulating risks, such as attempts at escape over multiple turns. Furthermore, customers can define their own session policies in straightforward terms.

• - Quarantine Mechanism: Upon detecting violations, the system can quarantine or halt the ID or session for a customer-specified duration, while administrators have the control to lift, extend, or make the measure permanent.

• - Human-in-the-Loop Approval: Instead of immediately blocking flagged operations, the system routes them through a human verification process for review and approval.

• - Comprehensive Audit Trail: To support security, governance, and compliance teams, Atlas logs all actions taken, complete with records of prompts, responses, and tool executions.

Agent IBAC effectively integrates with existing agents and AI tools that organizations already utilize, including Claude Code, Cursor, GitHub Copilot, and Microsoft Copilot Studio. This feature is available immediately to Varonis Atlas customers, representing a holistic approach to protecting AI agents from the development phase to runtime.

Additional Resources

For more detailed insights about Agent IBAC, explore our blog here. You can also request a demo of Varonis Atlas to see its operations in action.

About Varonis

Varonis (Nasdaq: VRNS) is dedicated to safeguarding data for secure AI utilization. The Varonis platform provides organizations with automated visibility and control features to secure critical data, ensuring a reliable AI environment throughout the entire process, from code to execution. With its 24/7 managed detection and response services, Varonis assists thousands of organizations worldwide in deploying AI, eliminating data leaks, and thwarting AI-driven threats.