• Trust3 AI secures agents on the Databricks Unity AI Gateway, controlling both what an agent can do and what data it can reach.
• Coverage includes agents built in Agent Bricks and outside agents connected through the gateway endpoint, with build-time controls enforced at runtime and every decision logged.
• The same controls run across the Azure AI Gateway, Cortex AI Gateway, and Bedrock AgentCore Gateway, with enforcement at each gateway and one policy across all of them.

SAN FRANCISCO: #AIAgents--Trust3 AI, founded by the creators of Apache Ranger, today extended its agent security controls to the Databricks Unity AI Gateway. Security teams now have one place to set and enforce policy across every agent gateway they operate.

Every Platform Has Its Own Gateway

Companies putting agents into production govern them one platform at a time. Each platform ships a harness and gateway for its own traffic, so policy lives inside each platform and nowhere above them.

"Every platform is shipping a gateway for its own agents and traffic, and that is the right thing for a platform to do," said Balaji Ganesan, Co-founder and CEO of Trust3 AI. "But no enterprise runs just one platform. Your agents cross Databricks and a number of other systems. We built the security control plane that sits above the platforms so unified enterprise policies can be reconciled and enforced across all of them."

Fragmentation costs enterprises in three ways. Agents stall in review, because security cannot see across platforms to determine what an agent will touch. Spend is metered gateway by gateway, so the total is unavailable until finance adds it up. Audits become reconstruction projects, because the agent’s action record is scattered across platforms.

A Gateway Does Not Check the Context

The security problem sits one layer down. A gateway asks whether a request is allowed:

Is the agent authenticated? Is the endpoint permitted? Is the call within its rate limit?

An agent can pass all three and still reach data outside the scope of its task, because the gateway evaluates the request without the context behind it.

A support agent querying a customer table looks identical to a fraud agent querying it.

What the Trust3 AI and Databricks Integration Delivers

The Unity AI Gateway governs AI traffic inside Databricks: agent activity, model requests, tool calls, and inference logging. Trust3 AI sits above it and adds:

• Purpose-Based Access Control: Trust3 AI builds policies from an agent's declared purpose and context, then enforces them inside the Unity AI Gateway through gateway plugins.
• Full Visibility and Monitoring: Trust3 AI discovers and profiles every agent configured on the gateway, and collects gateway traces to monitor token budgets, scope drift, and other signals.
• Continuous Posture and Runtime Enforcement: Trust3 AI continuously assesses each agent's security posture and expresses it as a Trust Score. Build-time configuration factors into the score, which informs what the agent is permitted to do in the Unity AI Gateway.
• Audit-Ready Records: Every agent decision leaves a record an auditor can retrieve on request.

"We weren't putting agents anywhere near sensitive client data until we could answer two questions: what is this agent allowed to do, and what data did it touch," said the Chief Information Security Officer of a global investment management firm. "Trust3 AI gave us the answer to both, at the gateway, which acts as the chokepoint."

The Same Controls on Every Platform

Databricks is one instance of a broader pattern. Trust3 AI brings a unified metadata and policy graph and applies the same controls across the Azure AI Gateway, Cortex AI Gateway, and Bedrock AgentCore Gateway. The platform extends to the agent frameworks and coding clients enterprises run alongside them. Enforcement happens at each gateway. The policy stays consistent across the board.

About Trust3 AI

Trust3 AI is the security control plane for enterprise AI agents and data. The platform applies purpose-based access control across agent actions and the data agents reach, from build time to runtime. Trust3 AI connects to any agentic framework and governs agents across clouds, platforms, and AI gateways including Databricks, Microsoft, and Snowflake environments. Enterprises use it to discover every agent, observe every decision, and secure every action. Trust3 AI is a Databricks Built On Partner, a member of the NVIDIA Inception program, and a member of the Snowflake Startup Accelerator.

Learn more at trust3.ai or request a demo at trust3.ai/demo.

Fonte: Business Wire