MLS Leaders Say Listing Data Is Already Feeding AI, and Licensing Rules Haven't Caught Up

At the Council of MLS (CMLS) Open House conference, MLS executives and two housing economists kept returning to one point: the listing data inside an MLS is its main asset, and AI has made the question of who can use it urgent. HousingWire reported on the sessions on October 1.

Why complete listing data matters

Danielle Hale, chief economist at Realtor.com, said a complete picture of the market is what lets sellers set a price and buyers decide what to offer. What happens when data goes missing depends on which part is missing. If a random slice is absent, the numbers hold up reasonably well. If the high end of the market is absent, reported prices come out lower than reality. If the low end is absent, the picture of the median seller and median sale price shifts.

Selma Hepp, chief economist at Cotality, added that fragmented data pushes cost onto consumers, agents, policymakers and researchers, who then spend more time checking whether information is current, complete and reliable.

Licensing was written for a different era

Several MLS leaders said current data licenses were built to distribute listings, not to govern AI. Colette Stevenson, CEO of REsides, described the typical license as one-dimensional. Ideas raised at the conference included tiered subscriber and data licenses that spell out which data streams a subscriber can access and what they may do with them.

Art Carter, CEO of California Regional MLS, suggested attaching metadata to data sets that states who owns the data and what counts as fair use. He said CRMLS is moving toward a set of use rules on every data package it sends out.

The AI question

According to the speakers, most MLS leaders and their broker members do not want listing data used to train large language models. Carter argued that a blanket ban is a mistake and unrealistic, because agents already use AI. In his view, MLSs should control the entry points through which AI reaches the data.

Chris Carrillo, CEO of North Texas Real Estate Information Services, warned that if MLSs stay out, their data will be used without agreements that protect brokers, and the revenue could go to another company instead of the brokers who supply the listings.

Andrew Coca, founder and CEO of Modern.tech and SourceRE, said "MLS data is already being used in AI." From his work at SourceRE, he said it is usually MLS members, not vendors, who go around the MLS, because the MLS has not given them a compliant way to use the data in AI.

Speakers also noted that AI output depends on the quality of the input, so fragmented or inaccurate MLS data will produce weaker results.

What this means when you compare real estate AI tools

This section is our reading, not something the speakers said. Nothing was decided at the conference. The speakers argued for changes, and each MLS sets its own rules.

Lead Generation, Property Valuation, CRM & Workflow and Listing & Marketing tools often depend on listing data. Before you commit to one, ask the vendor where its listing data comes from, whether that source allows AI use, and whether your own data is used to train models. Stevenson advised MLSs to keep asking vendors questions because products change within weeks. The same applies to buyers: ask about the current data pipeline, not only the demo.

Sources:

• HousingWire: "MLS data is more valuable than ever. The industry has to decide how to protect it" (Oct. 1, 2026)