Feds and States Sue Residential Landlords over AI Rent Software

The DOJ and several state attorneys general are going after software that uses AI to help landlords set rent, alleging antitrust claims against the landlords who use the software.  The regulators allege that a common real estate software application allows the landlords to illegally share pricing information with one another and thus collude to set rents artificially high.  The defendants are large residential landlords based in markets all across the country.  The allegations stem from the software itself, including its AI pricing algorithm, as well as behavior such as participating in user conferences to share information about the application.  

WHY IT MATTERS

As far as I know, this is the first antitrust action of any significance in the AI space, and it goes right to the heart of AI.  In essence, these antitrust claims go after the fact that a pricing algorithm is explicitly adapting based on customer inputs and that algorithm then guides the pricing decisions of other industry members.  

The software provider has access to rent information from its landlord customers; that information becomes part of the training data used to shape the rent/pricing algorithm employed by the software.  The regulators allege that this training data (along with other active behavior by the landlords) created conditions whereby property management “shared sensitive information about rental prices and used algorithms to coordinate” regarding rent.  This inherent information-sharing feature of the software was then amplified by the actual, direct communication among and between the defendants, including via user groups sponsored by the software provider.  (It is common in such groups for customers to share best practices for using the software, use cases in which they have deployed it, etc.)

Even before AI, it was common for a software platform provider to collect usage data and “anonymize” it to improve the performance of the software for other customers, or to sell aggregated benchmarking statistical data gleaned from users of an application.  It has also long been common to invite users to participate in round tables to discuss how best to implement and use the platform in their business – again, this is a perk that was provided long before AI.  What AI does, presumably, is improve the intelligence-sharing and make more of it real-time.  Where, as here, the users are generally all members of the same industry, we may see that regulators increasingly object to price-tweaking based on an algorithm that is trained on price data across an industry, especially if that industry is one ripe for consumer protection claims.  In this case, the industry users were not only using the software but were also actively sharing information about their use of the software and their business practices, where there may be even greater sensitivity.  

We do not know whether the incoming Trump administration will see this action as an important one; however, at least ten states have joined the case, so it will survive in some form regardless of whether the DOJ pursues it.  For the time being, this remains a case to watch.  And for any businesses buying tools that use industry pricing data to help set prices, this new antitrust action stands out as a potential cautionary tale.