MLS leaders say AI requires a new approach to governing real estate data

As artificial intelligence changes how real estate data is accessed, analyzed and deployed, two of the nation’s largest MLS leaders say the industry needs to rethink the infrastructure governing that data — and give brokerages far more visibility into how their information is being used.

During a discussion at HousingWire’s AI Summit, NorthStar MLS CEO Tim Dain and California Regional MLS CEO Art Carter outlined work underway to create a more granular data-governance model designed for an AI-driven industry.

Governance does not mean total control

For Dain, the distinction between governance and control is important. “I don’t like the word control because I think if any of you work with MLS data, which I’m assuming most of you do at some level, I think a lot of people look at us with a little bit of hatred because we probably force too much control and too many rules onto the industry,” Dain said.

Instead, he said the industry needs infrastructure that can identify who is using data, for what purpose, where it is being displayed and under what terms.

“It’s not really about the MLS controlling it as much as it’s about building an orchestration layer based on the proper entitlements, the proper authentications and the proper utilization and deployment of the data,” Dain said.

Carter said the issue has become increasingly important as brokerages seek greater authority over the data they contribute to MLS systems.

“So much of it is noise, and that noise can be pretty distracting,” Carter said of the debates taking place across the industry. “But really, the basis for all of that noise is really about brokers wanting to have greater control of their data and greater granularity of where it goes, how it goes there and, you know, what elements are going out the door.”

Most MLSs don’t currently have the infrastructure to provide that level of granularity, Carter said. CRMLS and NorthStar MLS have been working together on the issue for roughly 16 months.

Moving beyond traditional access controls

Dain said the effort involves moving beyond traditional role-based access systems and toward a model in which entitlements can be established at the field and record level.

That becomes particularly important as AI agents increasingly act on behalf of individual users. “Everybody’s going to have their own AI agent whether they know it or not,” Dain said. “If you bought one of these, it’ll probably just be built in and it’ll be transparent to you, but you’ll have an agent acting on your behalf.”

Carter said AI has accelerated the need for MLSs to recognize themselves as data companies. “MLSs are fast coming to the realization that we’re data companies,” Carter said. “And that realization comes with some responsibilities on how we manage that data.”

Traditional MLS policies were designed to apply broadly across participants, he said, but that approach may not provide enough flexibility for a market in which brokerages have different business models, technology strategies and approaches to AI.

“Policy, when it’s made, has got to be broad enough to handle everyone,” Carter said. “And that’s not really going to handle things in the new world with AI.”

For Dain, the answer is a policy engine capable of incorporating federal requirements, such as fair housing laws, state statutes, MLS rules and brokerage-specific business rules — and making those policies machine-readable and machine-enforceable.

Such a system could also make it easier for brokerages to authorize data access for mortgage companies, title companies and other partners rather than relying on complicated agreements governing each relationship.

“The entire ecosystem has to function off of that data,” Dain said. “So the brokerage needs the right abilities to grant the right entitlements to the right partners that it uses.”

Keeping brokers in the cooperative

Carter sees another risk if the industry doesn’t solve the governance problem: Brokerages could become less willing to contribute their data to the MLS ecosystem.

“MLS data is the oil that helps this industry run,” Carter said. “And there’s this real threat of the brokerage community in many cases taking their ball and going home.”

Maintaining the cooperative model is critical because MLS data ultimately feeds far more than real estate search, Carter said. It plays a role across the housing ecosystem, including property valuation and mortgage.

“Keeping that cooperative going and keeping as much of the data as possible in that cooperative for dissemination out to all the different elements in the industry is very, very important,” Carter said.

Dain believes a more sophisticated governance system could eventually change the economics of that cooperative as well.

He described a potential “charge for extraction, reward for contribution” model that would track who contributes data to the MLS ecosystem and who extracts value from it.

Under one hypothetical model, a brokerage contributing a complete listing with broad distribution rights could receive a full credit, while a listing entered after closing solely for comparable-sale purposes might receive only partial credit.

The concept could ultimately extend beyond listing data to leads, buyer information and other datasets contributed by companies throughout the housing ecosystem.

“Your interactions are metered and your contributions are rewarded and your extraction is charged for equally across the board, not to overpenalize anybody,” Dain said. “But if you’re a mass extractor, you should pay a lot more so that we can funnel the money back to the contributors so that they’re incentivized properly to continue contributing the data.”

AI is already creating data-governance problems

For Carter, this isn’t a theoretical problem waiting for the next generation of AI technology. He recalled visiting a brokerage office where 10 out of roughly 50 people said they were already uploading MLS data into Anthropic’s Claude.

“They are uploading MLS data into Claude with no governance whatsoever,” Carter said.

Rather than trying to prevent brokers and agents from using these tools, Carter believes MLSs need to create an environment in which they can use them while protecting the underlying data. “I don’t have to control the sandbox, but I do need to make sure that I do provide those opportunities for our brokers and agents to successfully do what it is that they’re going to be doing in this new world,” Carter said.

Dain warned that failing to establish those guardrails could ultimately leave the real estate industry paying companies for intelligence generated from the industry’s own data. “We’re in a position that if we don’t do this, we’re going to be buying back the intelligence created from our own data,” he said.

The growing ability of AI to perform functions traditionally handled by licensed professionals adds another layer of urgency.

Carter said California’s Department of Real Estate has indicated that brokers are ultimately responsible for their use of AI, but questions remain about what happens when AI itself begins providing real estate advice.

“AI is increasingly inserting itself into licensed activity, probably on the mortgage, the origination side and on the real estate side,” Carter said. “And, you know, that’s just the gray area that most states have no clue how to deal with.”

Governance as a path to innovation

Despite those concerns, both executives framed better governance as a way to expand access to MLS data rather than restrict it.

Dain said a properly governed system could ultimately make data available to a much wider range of companies, developers and potentially consumers because permissions and usage rules could be enforced at the technology level. “Governance makes innovation safe enough to scale,” Dain said.

His advice for companies developing their own AI strategies is to begin with the data rather than the product.

“First understand what data is being accessed. What use case is that data producing? Should it be governed and controlled? And how do you deploy it safely?” Dain said. “Those are your first answers. And if you can’t answer those questions, then you don’t have an AI strategy yet.”

Carter said MLSs aren’t inherently opposed to expanding access to their data. The problem is that the industry’s current governance mechanisms weren’t designed for the speed, scale or complexity of AI.

“Believe it or not, the MLS industry does want to give you access to its data,” Carter said. “The problem is, is the only governance method we have right now is a piece of paper, and once it goes out the door, we don’t know what it’s doing going out the door.”

Solving that problem, he believes, could fundamentally change the relationship between MLSs, brokerages, technology companies and the rest of the housing industry.

“I think that once we figure this piece of it out and that deliverable through the large language models,” Carter said, “we’ll be in a great, great place for everybody in the industry.”

This article was written by Tracey Velt with the assistance of HousingWire Automation, then reviewed by a HousingWire editor before publication.