Freight Brokers, Montgomery v. Caribe, and the AI Liability Gap

When a truck crash causes serious harm and the motor carrier's insurance runs out, claimants often turn to the freight broker. The Supreme Court's May 2026 ruling in Montgomery v. Caribe Transport II makes it harder for brokers to get these claims thrown out early. Brokers who use AI to choose motor carriers may face the toughest questions.

Key takeaways for brokers

• Interstate negligent hiring claims against freight brokers can no longer be dismissed early on FAAAA preemption.
• There is still no defined standard for what reasonable carrier vetting looks like.
• If AI helped pick the carrier, discovery will ask how, and who checked it.
• Human review only helps if it is real review.

What did the Supreme Court decide in Montgomery v. Caribe?

On May 14, 2026, the US Supreme Court unanimously held in Montgomery v. Caribe Transport II, LLC that a state-law negligent hiring claim against a freight broker is not preempted by the Federal Aviation Administration Authorization Act of 1994 (FAAAA), because such a claim falls within the statute's safety exception.

This removes a defense that freight brokers, serving as intermediaries between shippers and motor carriers, have relied on for years in courts that accepted it. Essentially, if there was a trucking casualty claim, freight brokers used to argue that they could not be held liable for negligent hiring or selection of motor carriers because federal law preempted the state-law claim, so it was dismissed before anyone examined whether they were careless. This meant a freight broker could avoid liability for choosing a 'dangerous' motor carrier (e.g., one with a poor FMCSA safety rating or a poor safety history) that was reasonably likely to cause crashes or injuries.

The ruling covers interstate shipments. The Court left open whether claims tied to purely intrastate broker arrangements can still be preempted, since that part of the statute has no safety exception.

Brokers must now be able to show that they took reasonable care in selecting a carrier. However, as Lockton notes in a recent report, "Montgomery did not create a definitive carrier-vetting standard or a clear definition of reasonable vetting." Most motor carriers also have no FMCSA safety rating, so brokers have to vet them using other data, and many now use AI tools to do this.

What does Montgomery mean for freight brokers using AI to select carriers?

So if freight brokers are relying on AI-generated recommendations for carrier selection without a clear definition of what reasonable care looks like, can AI governance protect them in the event of an accident? One answer may be human review of AI outputs, but once a claim is filed, discovery will ask more detailed questions:

• What did the broker deploy the AI tool to do?
• Did a person review the AI's recommendation before the decision was approved?
• Were safety ratings, insurance status or inspection history weighed, ignored, or overridden?
• Could the broker explain the AI decision at all?

Human review on its own may not be enough, as a current class action against Cigna, a health insurer, illustrates. In Kisting-Leung v. Cigna Corp. (E.D. Cal.), the plaintiffs allege that reviewers spent "an average of 1.2 seconds 'reviewing' each request." Review that fast may not be enough to demonstrate reasonable care.

If a human accepts an AI output without scrutiny, the use of AI can become evidence of a breach of the duty of care, or of negligent reliance on a system known to make errors. Prosperio Group CEO Beth Carroll, speaking at a recent logistics M&A conference, has said that brokers will need people behind these decisions, with controls and compliance around the rules the tools use.

What should brokers ask logistics clients who use AI for carrier vetting?

1. What is the AI tool deployed to do?

Does it rank carriers, flag them, or make the pick? Is it a vendor tool or built in-house? A tool that recommends is a different exposure from one that books loads with no person in the loop.

2. How is human review documented?

Who signs off, how long it takes, and what a reviewer does when they disagree with the tool. A log of overrides may be stronger evidence of reasonable care than a policy that says review happens.

3. What data does the tool weigh, and what happens when it's missing?

Most carriers have no FMCSA safety rating. Ask what the tool uses instead, such as inspection history, insurance status and out-of-service rates, and whether gaps trigger an escalation.

4. How would the current insurance program respond?

Contingent auto and broker liability forms were not written with AI in mind. Check them for generative AI exclusions, or for silence on AI-assisted decisions.

How can insurance respond to AI-assisted carrier selection claims?

Insurance is the second question. Many brokers' liability policies were not written with AI in mind. This is where affirmative generative AI error cover may step in, depending on the AI systems scheduled on the policy and its wording.

For those of us advising freight brokers, many clients already use AI in carrier selection. The question is whether their AI governance is robust enough to be relied upon as a defense, and whether their insurance program clearly responds when an AI-assisted decision is at the center of a claim. Belt and suspenders is always better.

How often do you discuss with your logistics clients how much they rely on their AI systems' recommendations, and whether their insurance would respond if those recommendations were wrong?