The Pulse
Collibra Buys trail ML to Enforce AI Policies at Runtime
Collibra announced its acquisition of Munich-based trail ML on October 5, 2026. The deal adds tools for automated AI governance assessments and runtime controls that can block agent actions that violate company policies.

Collibra announced on October 5 that it had acquired trail ML, a Munich-based AI governance company whose software is designed to assess controls and block policy-breaking actions by AI agents. The deal adds runtime enforcement to Collibra’s governance platform, extending its stated capabilities beyond documenting rules to applying them where agents operate.
From evidence review to continuous assessment
Trail ML’s agents can examine information about an AI system, identify relevant requirements and controls, assess whether those controls are in place and effective, and flag gaps. The technology can support governance tied to the EU AI Act, ISO 42001 and the NIST AI Risk Management Framework, Collibra says.
The companies also describe assessments that can run again when supporting evidence changes. That approach is intended to replace reviews conducted at isolated points in time with a more frequently updated view of an organization’s governance and compliance posture. The announcement does not specify how often reassessments would run or what triggers them beyond changes in evidence.
Collibra’s platform provides context for governing data, models, applications and agents, while trail ML adds automation for analyzing evidence, identifying requirements and managing governance workflows. Trail ML was founded in 2023 by Anna Spitznagel, Nikolaus Pinger and Sven Hölzel.
Policies reach the place where agents act
The acquisition’s distinct runtime element is meant to address what happens when an agent interacts with enterprise systems and tools. Trail ML can enforce its policies where agents run, blocking actions that violate them before they happen. The announcement contrasts that with governance focused on what a model says: agents can also take actions, making permissions at execution time part of the proposed control.
Collibra co-founder and CEO Felix Van de Maele tied the deal to the workload facing governance, risk and compliance teams. “AI governance cannot become another manual process that slows organizations down as AI scales,” he said.
Anna Spitznagel, trail ML’s co-founder and CEO, said the company was created to make governance automated and operational. “By combining trail ML’s automation and runtime capabilities with Collibra’s enterprise context, governance and reach, we can help organizations move faster with AI, safely and at scale,” she said.
The announcement leaves implementation details open
The announcement outlines the intended capabilities but does not provide a purchase price, a customer rollout schedule or technical details about how trail ML’s runtime controls will connect to specific agent platforms. It also does not say when customers will receive the combined capabilities. Those omissions leave the timing and mechanics of policy enforcement in deployed systems unspecified.
For Collibra, the announced addition links two stages of governance: assessing which rules apply and applying controls during an agent’s operation. The practical test will be whether the combined system can consistently translate an organization’s requirements into controls that actually stop an agent from taking an unauthorized action.
Sources
- Original source · Press release Collibra collibra.com
Other pages read
- PR Newswire prnewswire.com
Related stories
- David Robinson Resigns From OpenAI, Faulting Its Safety Culture
- Ping An Bank Board Approves First AI Management Rules
- Meta Adds Training-Time Safeguards to Its AI Release Framework
- Wikimedia Finds Unauthorized Activity It Links to OpenAI Agents
- Meta and Sierra Set Out an Open Protocol for Personal AI Agents