ROSS Intelligence doesn't exist anymore. If you're searching for it now, you won't find a live product. What you will find is one of the most consequential legal battles in the short history of AI — a lawsuit that started in 2020, outlasted the company it targeted, and ended with a federal court ruling that is still being digested by everyone building AI tools on legal data.
Here's what actually happened, in order.
What ROSS Intelligence Was
ROSS was a legal research startup that launched around 2015. The pitch was simple: instead of searching Westlaw or Lexis with keywords, you could ask ROSS a question in plain English and it would find relevant case law for you. The technology was built on top of IBM Watson in its early years. For solo practitioners and smaller firms who couldn't afford Westlaw's pricing, it was genuinely interesting.
It raised venture funding, got press, and was used by a number of law firms. It wasn't a massive company — but it was a credible early entrant in AI-assisted legal research at a time when almost no one else was doing it seriously.
The Thomson Reuters Lawsuit
In May 2020, Thomson Reuters filed a lawsuit against ROSS Intelligence in federal court in Delaware. The core allegation: ROSS had used Westlaw's legal headnotes to train its AI model without authorization.
This needs some context. Legal headnotes are the short summaries at the top of court opinions — "the court held that…" type summaries. Judges don't write those. Westlaw's editors do. Thomson Reuters has held copyright on those headnotes for a long time, and their argument was that ROSS effectively took that editorial work product and used it to teach its AI how to understand legal questions.
ROSS disputed the claims. But the lawsuit created an immediate problem: investors got nervous. Funding dried up. And in January 2021 — less than a year after the lawsuit was filed — ROSS shut down operations entirely, citing the litigation as a central factor.
The case continued even after ROSS closed. Thomson Reuters pursued it through the courts for years. This is worth noting because it shaped the legal landscape for every AI company that has since trained a model on copyrighted text.
The 2025 Court Ruling
In early 2025, a federal judge in Delaware issued a ruling in Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc. The court found in favor of Thomson Reuters, holding that ROSS had infringed Thomson Reuters' copyright in the Westlaw headnotes. The fair use defense ROSS had raised did not hold up.
This ruling landed in the middle of a much broader conversation about AI training data and copyright. The ROSS case was one of the first of its kind to reach a substantive ruling, and the outcome matters beyond legal research. It establishes that editorial work product added to public legal documents can carry copyright protection — and that training an AI model on that work product without a license is infringement.
Whether you agree with the ruling or not, it is now precedent. And it will be cited in future cases involving AI training data across industries.
What This Means for Legal Research Tools Today
The practical effect of the ROSS case is that it raised the barrier to entry for any new legal research AI that wants to compete with Westlaw or Lexis. You can't just scrape the major platforms and train on their editorial content. That narrows the field considerably.
It also explains something about the current market: most of the credible AI legal research tools today are either owned by Thomson Reuters or LexisNexis directly, or they're built on public court data and primary sources rather than the annotated, headnote-enriched content that the major platforms produce. That's not a coincidence. The ROSS litigation made the risk of the latter approach very clear.
A few tools have navigated this by partnering with the incumbents rather than competing with them. Casetext, before its acquisition by Thomson Reuters, built on public sources. The eventual $650 million acquisition in 2023 suggested that Thomson Reuters found it easier to buy competition than litigate it a second time.
Is There a ROSS Successor?
Not directly. The company is gone, the product is gone, and the team dispersed. Some of the people who worked at ROSS went on to other legal tech companies. But there's no continuation of the ROSS product under a different name.
What exists now is a more mature market. The natural language query interface that ROSS pioneered — ask a question, get relevant cases — is now a standard feature in tools like Lexis+ AI, Westlaw's AI-assisted research, and others. In that sense, ROSS's approach was right. The company just didn't survive long enough to see it become the norm.
The Honest Takeaway
ROSS Intelligence was an early and genuinely innovative legal AI company that got caught in a lawsuit it couldn't survive financially. The legal outcome — a ruling in favor of Thomson Reuters on copyright grounds — has real consequences for how AI companies can use legal databases to train their models going forward.
If you're evaluating legal research tools today and wondering why the landscape looks the way it does — dominated by a small number of well-capitalized players — the ROSS story is a big part of the explanation. Our guide to choosing an AI legal research tool in 2026 covers the current options and what distinguishes them.
20+ tools reviewed. No affiliate relationships with Thomson Reuters or LexisNexis.
Editorial note: AI For Legal Research publishes independent content. We do not accept payment for editorial coverage or review scores. Nothing on this site constitutes legal advice. Always consult a qualified attorney for legal matters.