For decades, the multiple listing service model has relied on a simple exchange: brokers contribute data, and in return, they gain access to a shared pool of market intelligence that benefits the entire cooperative. That exchange assumed a limited set of uses, mainly search and display, and a limited set of users. Artificial intelligence has changed both.
As AI companies build proprietary models from uncontrolled real estate data, multiple listing services risk permanently losing authority over the value their contributors have spent decades creating.
The Replication Problem
Every time a multiple listing service fulfills a data request, it creates another copy of its database. According to Tim Dain, President and CEO of NorthstarMLS, that replication model worked in a pre-AI world. It has become a structural liability today. “Right now MLSs are making 500 to 1,000 copies of their database every time they fulfill a data feed,” Dain says. “That was fine in a pre-AI world. It’s not fine anymore because that data, because there’s a copy that exists, the controller of that copy can decide what sort of intelligence gets created off of it.”
Each copy is a potential point of unauthorized AI training, derivative model creation, or intelligence generation. The original data contributors never approved this use and will never be paid for it. Dain argues the industry has already been living with ungoverned data extraction for years. AI is compounding the problem at a speed and scale that makes the status quo untenable.
Losing the Neutral Center
The MLS has historically functioned as the neutral center of a cooperative marketplace. It is the infrastructure that lets competing brokers share data and strengthen each other’s businesses. That neutrality, Dain argues, is now at risk of being displaced by outside entities whose primary incentive is profit rather than stewardship.
“I think we’re at risk of losing that neutral center,” Dain says. “I think we’re at risk of losing the authority over access, AI training, derivative products, any ability to monetize that for the content creator.”
The more alarming scenario, in Dain’s view, is that the window for recovery may be closing. “Once data becomes a model – like an AI model or proprietary intelligence – that control may get to a point where it’s not recoverable,” he says. Dain warns that in that scenario, companies would be forced to use whatever standard an outsider happened to create.
Dain points to his attendance at the AI4 conference as a sign of how fast the landscape is shifting. Out of roughly 400 booths, he says he did not see a single founder under 30 years old. He offers this detail not to dismiss the participants, but to underscore how many companies are being built on data that real estate professionals contributed without any governance framework in place.
Resetting the Incentive Structure
Dain’s core argument is not that outside companies are acting in bad faith. It is that the incentive structure is misaligned, and that misalignment will deepen as AI accelerates extraction. “The tragedy of the commons exists because somebody will always want to take more than they give,” he says. “I think that’s fine, but I think you can reset the incentive structure.”
The current subscriber-based model treats a small broker checking listings the same way it treats a large technology company training models on millions of records. It does not account for the volume of extraction, the type of intelligence being created, or whether the entity extracting value is contributing anything back. Dain argues this model made sense when the primary concern was data display and search. It does not make sense when AI can generate exponentially more intelligence from the same data, faster and at greater scale than any human user.
The comparison he draws to OpenAI is direct. “OpenAI didn’t reward the content providers for anything,” Dain says. “It just went and learned everything. There were all those lawsuits around the books that it consumed, the copyright it consumed from photos, from images, from paintings. That’s what we’re trying to avoid – that early mistake they made.”
Dain’s clearest summary of the risk: “We could end up buying back the intelligence that our data creates.”
A Governance-Based Solution
One proposed fix is to route all data interactions through a governance layer before they reach any end user or AI system, rather than allowing data to be copied out freely with each request. This layer would handle policy enforcement, entitlement controls, metering, audit trails, and compliance, making data movement traceable and permissioned instead of untracked. NorthstarMLS’s Nexus Re project is one example of this approach in development.
“We just shouldn’t allow AI to train on the data that brokers contribute,” Dain says. “It’s really hard to do though in an area where you have a thousand copies of your database out there.”
The broader shift this points to is a move away from flat subscription pricing toward a model where usage is metered and priced according to how much value is being extracted. Under this kind of approach, contributors who add value to the commons would pay less than those who only take from it. The principle behind it: those who take more from the system should pay more, and those who contribute should be rewarded for doing so.
About the Expert: Tim Dain is President and CEO of NorthstarMLS, the regional Multiple Listing Service of Minnesota, and leads its Nexus RE data governance infrastructure project.
This article is intended for informational purposes only and does not constitute legal, financial, or investment advice. The views and opinions expressed herein reflect those of the individuals quoted and do not represent an endorsement of any company, product, or service mentioned. Readers should conduct their own due diligence and consult qualified professionals before making any investment decisions.
