Don’t Get in Trouble Again, C.H. Robinson: AI Everywhere Except Carrier Vetting Is a Problem

(The views expressed here are solely those of the author and do not necessarily represent the views of FreightWaves or its affiliates.) C. H. Robinson is having an AI moment.

Its CEO, Dave Bozeman, has been publicly promoting the company’s Lean AI transformation, including AI agents, automation, productivity gains, appointment scheduling, quote responses, load tracking, and the broader message that C. H. Robinson is using technology to solve logistics problems at scale. That may be impressive.

It may be good business, and it may help C. H. Robinson move freight faster, improve margins, increase productivity, and serve customers more efficiently. But after Montgomery, the industry should be asking a harder question: is that same AI investment, data discipline, and human judgment being applied to carrier vetting? Because if C. H.

Robinson, or any digital broker, is building AI everywhere except the part of the business that decides which motor carriers are safe, legitimate, and suitable to put on the road, that is not innovation. That is exposure. The timing is hard to ignore. C. H. Robinson just lost a unanimous Supreme Court case involving broker negligent selection.

The claim was that C. H. Robinson negligently hired a motor carrier with a conditional safety rating and that the broker knew or should have known from that rating that selecting the carrier was reasonably likely to result in crashes that would injure others. Now, almost immediately after that decision, C. H.

Robinson is publicly emphasizing the sophistication of its AI transformation. That creates a fair question for the industry: if the company has the capital, data, systems, and technical discipline to automate and optimize so many parts of freight brokerage, why should carrier vetting remain the place where sophistication stops?

The Supreme Court did not say carrier vetting must be manual. It did not tell brokers to stop using technology, and it did not require brokers to become motor carriers, inspect trucks, audit driver qualification files, or supervise drivers.

But the Court did make clear that negligent hiring claims against brokers are not categorically preempted when they involve motor vehicle safety, and the majority opinion framed the issue around ordinary care in selecting a carrier. Justice Kavanaugh’s concurrence is especially important for digital brokers.

He acknowledged that brokers may not always be in a position to objectively assess the relative safety of every motor carrier, but he also wrote that brokers may sometimes become aware that a particular carrier operates unsafe trucks or hires unfit drivers.

He emphasized that brokers should be able to defend themselves when they act reasonably and arrange transportation with reputable carriers, and he quoted plaintiff’s counsel saying brokers should hire carriers with a reasonable policy and ask the hard questions of the carrier. That language matters.

It does not describe carrier vetting as a blind box checking exercise. It describes judgment, inquiry, escalation, and a reasonable response to information when the broker has it or should have had it. This is where AI alone can fall short. Carrier vetting has a human component because not every carrier with a risk indicator should be automatically rejected.

If a broker tries to fully automate carrier vetting with rigid pass or fail rules, it may lose significant capacity and still miss the practical judgment required to understand whether a carrier is appropriate for a particular shipment. The human element is not ignoring the data.

The human element is asking the hard questions when the data shows something that matters. It is someone looking at the red flags, digging deeper, deciding whether the issue is disqualifying or manageable, documenting the reasoning, and applying appropriate controls if the carrier is still used.

Why does this carrier have no inspections despite its claimed operations? Why is the authority so new? Why does the contact information not match FMCSA records? Why is the carrier’s volume inconsistent with its apparent fleet size? Why are there related entities with concerning histories? Why is the payment information inconsistent?

Why is there a carrier identity mismatch at pickup? Why does this carrier appear suitable for this shipment despite a risk indicator? Those are not questions an algorithm should hide. Those are questions technology should surface so the right person can make a documented and defensible decision.

I recently interviewed Michael Leizerman, the plaintiff attorney who won Montgomery at the Supreme Court. During that interview, I asked him directly about brokers using AI and large data sets to automate operations and assign carriers.

His answer should concern every digital broker because he made a critical distinction between AI that improves safety and AI that merely increases speed. Michael recognized that AI can be used to make the roads safer. It can help identify chameleon indicators, r