Zentera Secures AI Agent Use of Sensitive Data with New Data Access Control Features

Ensage AI Security now gives security teams controls, so companies can adopt agents without exposing licensed designs, research data, or customer records.

Milpitas, CA (PRUnderground) September 29th, 2026

Zentera Systems today announced Data Access Control, a new capability in Ensage AI Security. An AI agent usually runs with the file permissions of the person who started it. Data Access Control gives security teams the ability to govern which files each agent can open. Engineers keep their current access, and agents get only the access that the security team approves. Ensage AI Security supports Data Access Control from release 10.2.

Security leaders see agent access to data as an open gap. In Agents of Change, a Zentera survey of 251 cybersecurity leaders in semiconductor, pharmaceutical, and financial services organizations, 80 percent said they are concerned that AI agents in their environment may hold access that was never explicitly granted. Another 84 percent said agents can cross project boundaries more easily than employees. An AI agent can send any file it can open to a model provider, a file transfer service, or other external services.

With Data Access Control, security teams assign file and folder access permissions to agents. Ensage AI then enforces each decision on the machine that runs the agent, before the agent reads any content. Policies cover files and folders on local and network storage, and permissions are inherited by the tools the agent calls.

For example, a semiconductor design team can let AI agents run the approved EDA tools that use a foundry PDK, while the agents cannot open the PDK files in any other way. The team applies the limits in foundry license agreement to AI agents, and its engineers work as before.

“Security teams have to account for every path an AI agent can use to move data out of the environment,” said Jaushin Lee, President and CEO of Zentera Systems. “Until now, most controls have watched traffic at a gateway, and a gateway sees only part of what an agent can do. Data Access Control puts the decision on the file itself. An agent cannot leak the contents of files that it cannot open. That lets our customers put AI agents to work on projects that involve restricted data.”

Availability
Data Access Control will be available in Ensage AI Security release 10.2.