MLS data is highly valuable, which is why it’s protected through licensing agreements. These agreements ensure compliance with copyright laws and govern who can display listings, access data and use data for analytics.
With the rapid rise of artificial intelligence tools, however, that valuable data could be at risk of being copied, ingested, analyzed or repurposed by AI systems, which are outside the scope of original data licenses.
For Chris Lambrou, CEO at Milwaukee’s Metro MLS, that risk was the impetus for its new AI policy. Metro MLS is the largest multiple listing service in Wisconsin and serves more than 9,000 members at six local REALTOR® associations: Greater Milwaukee Association of REALTORS®, Jefferson County Board of REALTORS®, Lakes Area REALTORS® Association, Lakeshore REALTORS® Association, Manitowoc County Board of REALTORS® and Southshore REALTORS® Association.
Policy Matters
In general, Metro MLS’ AI policy is simple and direct:
“Metro MLS data may not be used for AI training or machine learning in any way without Metro MLS’s express written permission. This includes AI-related derivative datasets and AI processing of MLS data for predictive or generative capabilities.”
“If someone violates this section, it is considered a material breach of the agreement, meaning Metro MLS can pursue contract remedies and any other legal remedies available,” Lambrou says.
To clarify understanding, Metro MLS’ policy also defines exactly what it means by AI and training:
- AI is artificial intelligence systems, machine-learning models, neural networks, large language models, generative AI, predictive models, automated decisioning systems, recommendation engines, computer vision models or similar technologies.
- Training is fine-tuning, testing or validating AI models; improving model outputs; creating embeddings or vector databases; or creating synthetic or derivative datasets.
“As information sources for the industry, we need to continue our practice of protecting (against) the misuse or abuse of our data compilations,” Lambrou says. “There are so many unknown and endless possibilities in the realm of AI, so a less-is-more approach is paramount to paying it forward for the near and distant future.”
Of course, repurposing MLS data is always prohibited and applies here as well, he adds.
Policy Developments
The policy was developed during Metro MLS’ broader review and update of its data-licensing agreements. During that process, the service’s management team, contract-management support and legal counsel reviewed existing data-license language, identified AI-related data-use risks and added language to clearly prohibit AI use without prior written authorization.
“The goal was to address emerging risks around artificial intelligence and machine learning, specifically the possibility that MLS data could be used to train, model and support AI-related products beyond the intended scope of the license,” Lambrou says.
Policy in Practice
Users have welcomed the policy, recognizing the need for guidance and guardrails. “What I’ve found is that the more serious AI vendors we encounter already get it,” Lambrou says. “They already do not comingle MLS complications with AI. They already have strict data-privacy practices in place from the ground up.”
As AI tools continue to become more adept, Lambrou adds that protecting MLS data is too important to leave to static policy update workflows. “Being ever-present and available to our subscribers helps us stay in the conversation of their needs, and being technology-focused keeps us on the front page of progress and manipulation in AI,” he says. “Timing is too important here. The aggressive growth of AI technology and the never-ending quest for improved workflow of the real estate practitioner make this an ongoing, 360-degree task.”
An AI policy is just the first step. As we all test the boundaries of what AI can do, Lambrou suggests continuing to educate industry members on how to use AI within the recommended practices of the National Association of REALTORS®, MLSs and state regulatory agencies.
For example, having an AI assistant act as an agent or member of a team may skirt state laws. AI cannot give guidance on transactions, including answering questions like, “Is this home overpriced?” or “Should I offer below asking?” or “How competitive is this neighborhood?”
“Regulators could argue it’s performing a brokerage service, which could violate state licensing laws,” Lambrou says.
Other issues on the borderline include using AI to rewrite MLS listing data, to re-rank listings, generate property images, write reviews or testimonials, scrape leads from MLS or Internet Data Exchange sites, or coordinate showings.
“AI is a remarkable disruption for our industry,” Lambrou says. “It’s like a shade going up in a dark room … suddenly, everything is lit. As leaders, our responsibility is to make sure that light is directed in the right places, with the right safeguards and for the benefit of the marketplace we serve.”
