Databricks Bolsters AI Governance

Databricks is doubling down on AI governance with significant updates to its Unity AI Gateway, unveiled at the Data + AI Summit 2026. The expanded capabilities aim to provide enterprises with granular control over increasingly complex AI deployments.

The core of the announcement revolves around extending Unity Catalog's governance framework beyond data to encompass AI assets and their runtime interactions. This move addresses the growing challenge of managing multi-model, multi-agent, and multi-vendor AI estates.

Optimizing AI Spend and Usage

Organizations can now gain detailed visibility into AI expenditures across various tools and models. Features include unified spend tracking, granular cost attribution by user or team, and hard spend caps to prevent runaway expenses.

Intelligent routing capabilities help balance model quality with cost efficiency, directing workloads to the most appropriate resources based on task complexity and budget constraints.

Unified Governance for AI Assets and Interactions

Unity Catalog now registers and secures not just models, but also Managed Cloud Provider (MCP) services, agents, and skills. This unified approach provides centralized visibility and auditability for all AI components.

The platform enforces runtime controls through Contextual Service Policies, allowing administrators to define specific actions and guardrails for AI agents. These policies can be applied based on user, agent, model, or the content of requests and responses.

Databricks is also offering managed MCP services for popular enterprise applications like Google Drive, Jira, and Slack, simplifying integration and governance.

Enhanced Monitoring and Security

End-to-end agent tracing provides a unified telemetry layer to monitor model interactions and tool activity. This allows for easier troubleshooting and investigation of AI workflows.

Coding agent activity can be analyzed using natural language queries, identifying inefficiencies and opportunities for productivity gains. Security investigations are streamlined through Lakewatch, enabling faster detection of suspicious activity and policy violations.

Extending Governance Through an Open Ecosystem

Unity AI Gateway is designed as an open platform, integrating with a growing ecosystem of AI security, identity governance, and threat detection partners. Upcoming integrations will include major players in AI security and identity management.

This approach ensures that organizations can bring trusted controls into their runtime AI workflows, leveraging existing security investments. The Unity AI Gateway is positioned as the foundation for enterprise AI governance.