The Antitrust Risk Hiding Inside Your AI Pricing Software

Dynamic pricing software has been sold for years on a simple pitch: stop guessing, let an algorithm read demand and set the right price in real time. Hotels, self-storage operators, car rental companies, and a growing list of retailers use some version of it, usually bought from a third-party vendor rather than built in-house. That category of software is now at the center of a live antitrust fight, and the outcome matters to any business that has adopted one of these tools without thinking much about where its pricing data actually goes.

The legal theory getting traction is not new in concept, but it has a new mechanism. A "hub-and-spoke" conspiracy is one where competitors don't coordinate directly with each other — they each deal separately with a common vendor, the hub, who uses what one competitor tells it to influence what it tells the others. Antitrust law has treated that as functionally equivalent to a horizontal agreement for decades. What's changed is that a shared pricing algorithm can now perform that coordination automatically, continuously, and without any of the participating businesses ever picking up a phone. In November 2025, the Department of Justice reached a settlement with RealPage, a vendor whose software pools confidential rent and occupancy data from competing landlords to generate pricing recommendations, restricting how that pooled data can be shared going forward. In July 2026, the Third Circuit revived a similar claim against a group of Atlantic City casino-hotels, ruling that allegations they used a common algorithmic pricing vendor to set room rates were plausible enough to proceed, even without proof of who said what to whom.

That doesn't mean using a shared vendor is automatically illegal. A federal appeals court held the opposite in Gibson v. Cendyn Group in 2025: merely subscribing to the same pricing software as a competitor is not, by itself, evidence of collusion. But the same ruling was explicit that this is not a safe harbor — it's a fact-specific question about what the algorithm actually does with each customer's data and how much independent judgment the business retains over the price it ultimately charges. "We just use the same software everyone else does" is a defense that depends entirely on details most buyers never ask about at the point of purchase.

The stakes moved from civil to potentially criminal this year. In May 2026, a senior Justice Department antitrust official said publicly that the department views coordinated use of algorithmic pricing tools, shared SaaS platforms, and large language models as a route to criminal Sherman Act liability, not just the civil exposure these cases have carried historically. A business that adopted a pricing tool purely as a revenue-optimization decision may not have registered that the purchase now sits inside an area of active criminal enforcement interest.

None of this requires a business to have intended anything wrong. The exposure comes from the mechanics of the tool, not the motive for buying it. A pricing platform that ingests a subscriber's current rates, occupancy, or inventory levels and factors that data into the recommendation it generates for other subscribers in the same market is doing, mechanically, close to what a hub-and-spoke conspiracy requires — regardless of whether the business using it ever thought about a competitor's prices at all. A tool that only uses a customer's own historical data and public market signals is a materially different, and safer, product, even if the sales pitch sounds identical.

The practical response is a short, direct conversation with the vendor, ideally before signing or renewing rather than after a complaint arrives. Ask plainly whether the pricing recommendation your business receives is influenced by non-public data submitted by other subscribers, including competitors, and ask the same question in reverse — whether your own data feeds recommendations made to them. Ask whether pricing is applied automatically or surfaced as a recommendation you can override, and if it's the former, ask what it would take to move to the latter. Get the answers in writing, since a sales call's assurances carry little weight if the arrangement is ever examined later. And keep a record, even an informal one, of the instances where your business set a price that departed from the tool's suggestion — that record is the clearest evidence of genuine independent pricing judgment if the question is ever raised.

None of this is a reason to abandon dynamic pricing, which remains a legitimate and often significant efficiency gain over manual rate-setting. It's a reason to evaluate a shared pricing algorithm the way a careful business would evaluate any arrangement that touches a competitor's data — with specific questions about data flow and independent judgment, asked at procurement, rather than assumed because the software is popular and everyone in the industry seems to use it. The businesses least exposed here won't be the ones that avoided AI pricing tools. They'll be the ones that knew exactly what their vendor's algorithm was doing with their numbers before they let it set a price.