In February 2025, a federal court handed down a ruling every MLS executive and broker should read. Thomson Reuters, the owner of Westlaw, sued an AI startup called Ross Intelligence for training its legal search engine on Westlaw’s headnotes. Those headnotes are short editorial summaries of judicial opinions, written by attorney-editors. The underlying court opinions are public domain. The headnotes are not.
Here is what happened. Ross wanted to build an AI-powered legal research tool that competed directly with Westlaw. It asked Thomson Reuters for a license to the headnotes and was refused, because Ross was a competitor. So Ross commissioned roughly 25,000 “Bulk Memos” from a third-party contractor, each memo pairing a legal question with answers drawn from the Westlaw headnotes, and trained its AI on those memos. The headnotes never appeared in Ross’s product. They were the invisible training fuel behind it.
Judge Stephanos Bibas, sitting in the District of Delaware, granted summary judgment for Thomson Reuters. The use was not transformative, the court held, because Ross was using the headnotes for exactly the same purpose Westlaw uses them: powering an efficient legal research tool. The court distinguished the famous intermediate-copying cases like Google and Sony, noting that Ross did not need to copy the headnotes to reach the underlying legal ideas, which were freely available in the public court opinions. Ross had even asked Thomson Reuters for a license first, been refused because it was a competitor, and then obtained the training material through a third party. The court called the dispute, in essence, an ordinary copyright case. The ruling is now on appeal to the Third Circuit, but the trial court’s message could not be clearer.
Now read that again and substitute the words “property descriptions” for “headnotes” and “property search” for “legal search.” The parallel is exact, and the real estate industry should be paying attention.
Listing data is editorial work
Nobody disputes that the raw facts of a transaction are not copyrightable. A house exists at an address. It sold for a price on a date. Those are facts, and facts belong to everyone.
But an MLS listing is not a pile of facts. It is a curated work. An agent writes a property description that frames the home, chooses which features to emphasize, and crafts language designed to attract buyers. A photographer composes shots, selects angles, edits for light. The MLS compiles all of this into an organized database with fields, taxonomies, and search structures that took decades to build. That is editorial labor, the same kind of labor the court protected in Westlaw’s headnotes.
The Westlaw case turned on exactly this distinction. Judicial opinions are public, but the attorney-editor’s summary of the opinion is original creative work. In real estate, the underlying facts of a sale are public, but the listing as assembled, described, and presented is the product of professional judgment. When an AI company scrapes that data to train a competing property search tool, it is not harvesting facts. It is exploiting original content.
The fair use argument fails the same way
Fair use is judged on four factors, and the Westlaw ruling is instructive on the two that matter most here.
The first is the purpose and character of the use. Ross argued its AI training was transformative. The court disagreed: Ross was using the headnotes to build a legal research tool, which is precisely what Westlaw is. There was no new purpose, no different character. An AI company that scrapes MLS data to train a competing home search product is in the same position. It is not transforming the listings into something new. It is rebuilding the same product without paying for the content.
The fourth factor is the effect on the market for the original work. This is where the damage is most direct. A scraped training set lets a competitor replicate the value of the listing database without the cost of building it: no agent relationships, no data feeds, no quality control, no decades of compilation. The court recognized that this kind of market substitution is exactly what copyright is meant to prevent.
I wrote about the fix last week
Last week I published a companion piece on the indexing problem: why only the listing broker should decide what gets indexed. Read it here: https://www.wavgroup.com/2026/10/08/the-indexing-problem-why-only-the-listing-broker-should-decide-what-gets-indexed/
The argument is simple. For more than twenty years, every listing photo, description, price change, and sold record has been crawled, stored, and repurposed by search engines and portals that spend nothing to acquire the data and carry none of the liability for it. If there is a square footage error or a fair housing problem in the remarks, the search engine has no liability. The listing broker does. The value of the listing asset moved to the indexer. The liability stayed with the broker.
The fix is already in the code. The listing broker’s own site carries index, follow. Every cooperating broker’s IDX or VOW display of that same listing carries noindex, nofollow, with the AI training crawlers (GPTBot, ClaudeBot, Google-Extended, CCBot) disallowed in robots.txt. The search engine finds the listing once, from the authoritative source: the listing broker. Displaying a listing under IDX is not the same as authorizing its indexing, and the listing broker, as the seller’s agent, should hold the exclusive right to decide if, how, and where the listing is indexed. One listing. One broker. One decision.
The Thomson Reuters ruling is the legal foundation underneath that technical fix. The indexing article is the implementation. Together they give the industry both the right and the mechanism.
What brokers and MLSs should do
The ruling is good news, but rights you do not enforce are rights you lose. Protecting listing data requires deliberate action on two fronts.
The first is contracts. Every data feed, every API agreement, every vendor relationship should carry clear licensing terms that define what the data may be used for and, just as importantly, what it may not be used for. AI training should be addressed explicitly, not left to implication. Westlaw protects its intellectual property with robust licensing, and it had the contracts to stand on when it went to court. Brokers and MLSs need the same discipline. If your agreements are silent on AI training, fix them now.
The second is technological measures. Licensing terms mean little if the data is freely scraped by anyone with a crawler. MLSs should enforce their terms with the tools available: access controls on data feeds, monitoring for unauthorized bulk access, and the noindex and robots.txt controls described above. The Thomson Reuters ruling strengthens the legal position considerably. The indexing controls supply the technical one.
The asset of the future, protected by the law of the past
I have been arguing that software is dying and data is the asset of the future. The Westlaw ruling is the legal system starting to agree. The interface can be rebuilt by anyone with agents and a spec. The data underneath, the curated, compiled, editorially shaped listing data that the industry produces every day, is original work with real copyright protection behind it.
That protection only matters if the industry treats its data like the asset it is. License it clearly. Guard it technically. Enforce it when someone takes it. The brokers and MLSs that do this will own the most defensible asset in real estate’s AI future. The ones that leave their data unguarded will watch someone else build a business on it.
The post Your Listing Data Is Copyrighted. Start Acting Like It. appeared first on WAV Group Consulting.
Leave a Reply