Unlocking Agentic AI: Immuta and Databricks Target Enterprise Security Gaps

SAN FRANCISCO, CA – June 15, 2026 – In a significant move to secure the next wave of enterprise innovation, data access governance leader Immuta today announced an expanded partnership with Databricks, introducing a suite of capabilities aimed directly at the complex security challenges posed by autonomous AI agents. The collaboration delivers a new framework for how companies can safely deploy and scale AI on the Databricks platform, moving beyond legacy security models to enable dynamic, intent-driven data access.

This partnership marks a critical milestone in the journey from AI prototypes to profitable, enterprise-wide systems. As organizations race to leverage autonomous agents for tasks ranging from data analysis to business automation, they are confronting a major bottleneck: data security. The new solutions from Immuta and Databricks are designed to dismantle this barrier, providing the governance infrastructure necessary to translate AI potential into commercial reality.

The High Stakes of Autonomous AI

The promise of autonomous AI agents—systems that can independently reason, plan, and execute tasks—is immense. However, their ability to interact with vast corporate data lakes also presents a formidable security risk. Traditional security models are ill-equipped for this new paradigm. Static service accounts often grant agents dangerously broad access, creating a massive surface for data exfiltration. Conversely, OAuth-based architectures, which require pre-provisioning accounts for every potential human user, create an operational nightmare for IT teams and introduce the risk of catastrophic privilege escalation if an agent inherits the permissions of a high-level executive.

"Deploying enterprise AI hits a breaking point when organizations try to force modern agents into legacy security frameworks," said Steve Touw, CTO at Immuta. "Relying on OAuth for AI agents forces IT to provision accounts for every potential user while risking catastrophic privilege escalation."

This security-versus-scalability dilemma has, until now, forced many companies to keep their most powerful AI initiatives locked in sandboxed environments, preventing them from delivering real business value. The challenge is clear: to unleash AI, enterprises must first be able to control it with precision and confidence.

A New Governance Model: Intent-Driven Access

The centerpiece of the announcement is a shift from basic authentication to a sophisticated, centralized authorization model. By deeply integrating with Databricks Unity Catalog—the platform's unified governance layer—Immuta introduces four key capabilities that work in concert to create a secure environment for AI agents.

Immuta Agentic Data Access is the core of this new model. It allows an AI agent to operate securely on behalf of a human user. Instead of using a generic, high-privilege token, Immuta validates the user's identity and dynamically bounds the agent's session to only the data that specific user is authorized to see—down to the individual row, column, or cell. This happens in real-time, with no need for pre-provisioned accounts and with a full audit trail maintained in Unity Catalog.

This is powered by Intent-Driven Access Control, a paradigm that grants permissions based on the verified context of a specific task. Access is provisioned just-in-time for the duration of the task and automatically revoked upon completion. This 'just-in-time, least privilege' approach minimizes risk exposure and ensures continuous compliance with regulations like GDPR and HIPAA. It effectively creates a 'double-lock' system, where an agent's access is limited by both its intended task and the user's underlying permissions, preventing unauthorized actions even in the face of risks like prompt injection.

As Touw explained, this approach "moves past basic authentication to true, centralized authorization, eliminating the provisioning bloat and ensuring access is always verified, tightly scoped, and secure."

From Governance to Growth: Enabling Commercial AI at Scale

Beyond securing agentic workflows, the partnership delivers tools that directly accelerate the path to commercialization. The Immuta "Comply" App for Databricks Unity Catalog transforms compliance from a manual, time-consuming audit into a dynamic, conversational query. Business and security teams can now ask plain-English questions like, "Which AI agents have access to financial tables?" and receive instant, audit-ready answers, dramatically reducing the friction in governance.

Furthermore, a new scalable architecture for Group-Based Permission Assignment ensures that as an organization's use of Databricks grows, its security policies can scale without hitting platform limits. By intelligently grouping users with similar permission sets, Immuta's engine bypasses internal limits on principal-to-object assignments, ensuring access automation remains robust at enterprise volumes.

This comprehensive approach is designed to help companies move AI projects out of the lab and into production faster. "Customers consistently ask us for simpler, more seamless ways to bring their data, analytics, and AI together—and to move from experimentation to production faster," noted Stephen Orban, SVP of Product Ecosystem & Partnerships at Databricks. "With this new integration, Immuta is helping address that need."

Charting a Course in a Competitive Market

The Immuta-Databricks announcement lands in a rapidly growing and fiercely competitive AI governance market. Projections show the market soaring from approximately $197.9 million in 2024 to over $6.6 billion by 2034, fueled by regulatory pressures and the urgent need to mitigate AI-related risks. Companies like Privacera, Alation, and Velotix are also vying for dominance in this space.

Immuta's strategy is to differentiate through deep, native integration within a leading data and AI ecosystem. By building directly on top of Databricks Unity Catalog and tailoring its solutions to the specific challenges of autonomous agents on that platform, Immuta is creating a powerful, specialized offering. This tight partnership provides joint customers with a solution that feels less like a bolted-on third-party tool and more like an organic extension of their existing data platform, a critical factor for driving adoption and ensuring long-term commercial viability.

Paving the Way for a Zero-Trust AI Future

Ultimately, this collaboration is about more than just a set of new features; it's about laying a foundational layer of trust for the future of enterprise AI. By providing a robust, scalable, and auditable framework for data access, Immuta and Databricks are de-risking the adoption of autonomous AI. This enables organizations to build with confidence, knowing that their most valuable data assets are protected even as AI agents are given more responsibility.

This verifiable control is the key that will unlock the next generation of AI-driven business transformation. With a zero-trust architecture in place for AI agents, companies can finally move from cautious experimentation to building sustainable, strategic AI capabilities that interact with data continuously and autonomously, truly reshaping business processes and creating lasting competitive advantage.