Collibra Buys trail ML to Give AI Its Own Automated Compliance Department
The acquisition brings agents that assess AI controls and can block policy-breaking actions. Supervising the automation is acquiring an automation problem of its own.
Automation ·
Somewhere in the career plan, "I'll supervise the AI" sounded like a sensible escape route. Collibra has acquired trail ML, whose specialty is putting agents to work on that supervision. An awkward development for the escape route.
The October 5 announcement describes software that examines evidence about AI systems and checks which requirements apply. It then assesses whether the relevant controls actually work. Collibra says changed evidence can trigger fresh assessments, reducing the manual effort involved in keeping compliance records current.
I admire the thoroughness. First you install an agent to do the work. Then you install another to establish whether the first was allowed to do it. Your expertise in asking "has this been checked?" now faces competition from software that remembers the answer.
There is a less decorative part of the deal: trail ML's runtime technology is designed to block actions that violate policy before an agent carries them out. That gives the automated compliance department something beyond a retrospective frown. I was rather counting on making excellent progress before anyone reached the section about permission.
Based on: Collibra acquires trail ML to automate AI governance from policy to production, Collibra, October 5, 2026.