Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
Thomson Reuters won a major U.S. copyright ruling against Ross Intelligence—but the decision is narrower than headlines suggest. On February 11, 2025, a Delaware federal judge ruled that Ross’s use of Westlaw-derived headnotes and classification material to develop a competing legal-research product was not fair use.
That is significant for companies building commercial AI products from proprietary databases. It is not, however, a blanket ruling that training ChatGPT-style models on copyrighted books, news, images, music, or code is automatically unlawful.
| # | Preview | Product | Price | |
|---|---|---|---|---|
| 1 |
|
Copyright Law | $166.82 | Buy on Amazon |
| 2 |
|
Copyright Law: Cases and Materials (v8.0) | $21.70 | Buy on Amazon |
| 3 |
|
Copyright Law of the United States: and Related Laws Contained in Title 17 of the United States Code | $10.32 | Buy on Amazon |
| 4 |
|
Copyright Law in a Nutshell | $65.00 | Buy on Amazon |
| 5 |
|
Copyright Handbook, The: What Every Writer Needs to Know | $37.99 | Buy on Amazon |
What Thomson Reuters actually won
The case is Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc. Thomson Reuters sued Ross in 2020, alleging that Ross used protected Westlaw material while developing an AI-powered competitor.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →Thomson Reuters operates Westlaw through its legal-information business. The dispute was about Westlaw—not Reuters news articles.
#1 Best Overall
On February 11, 2025, the district court granted Thomson Reuters partial summary judgment on important copyright-infringement issues, rejected Ross’s fair-use defense, and denied Ross’s own summary-judgment motion on Thomson Reuters’ copyright claims. It was a major liability ruling, but not necessarily a final judgment resolving every remedy, damages issue, or procedural step.
Ross Intelligence is now defunct. Its proposed product was designed to compete with Westlaw by answering legal questions and returning existing judicial opinions rather than generating entirely new prose in the manner of ChatGPT.
Read the federal case record and the February 11, 2025 opinion.
Free tools Windows power users keep installed
One-click scans. No signup required.
What material was at issue?
The dispute did not concern Thomson Reuters owning the law itself. Court opinions, legal rules, and underlying facts are different from the editorial work added by a legal publisher.
The contested Westlaw material included:
- Headnotes: attorney-written summaries identifying significant legal points in a judicial opinion.
- The West Key Number System: a proprietary system for classifying and organizing legal issues.
The court found that the selection, wording, and organization of these editorial elements could meet copyright’s originality threshold. The distinction matters: a company may not own the underlying legal holding, but its editorial expression and classification system can still receive protection.
Why Ross’s fair-use defense failed
Fair use is a fact-specific defense. The court’s analysis focused particularly on the commercial purpose of Ross’s use and its effect on Westlaw’s market.
| Fair-use factor | How it mattered in this case |
|---|---|
| Purpose and character | Ross was developing a commercial product that competed directly with Westlaw. The court found that Ross had not added enough new expression, meaning, or purpose to make its use sufficiently transformative. |
| Nature of the work | This factor was more favorable to Ross because the headnotes concerned legal material and judicial decisions. It did not outweigh the other considerations. |
| Amount used | The court did not view this factor as enough to overcome the problems created by the commercial and competitive use. |
| Market effect | This was especially damaging to Ross. Its product was intended as a substitute for the same legal-research service Westlaw provided. |
The practical lesson is not simply that “AI copied text.” The judge viewed the conduct as using a rival’s protected editorial work to build a product serving essentially the same market purpose.
The big asterisk: this was not a ChatGPT case
Ross’s system was a non-generative legal-search product. It returned existing judicial opinions in response to legal questions. That makes the case materially different from disputes involving general-purpose language models, image generators, music systems, or coding assistants.
Generative-AI cases can raise additional questions, including:
- Whether copying works into a training process is transformative.
- Whether the data was acquired lawfully or under a license.
- Whether a model stores or reproduces protected expression.
- Whether outputs reproduce memorized passages or other protected material.
- Whether the AI product competes with the copyright owner or its licensing market.
- Whether the plaintiff can prove access, copying, and measurable market harm.
A general-purpose model may not compete with a source publisher in the same direct way Ross competed with Westlaw. That difference could matter, but it does not guarantee fair use.
Rank #3
The safer description is: a Delaware judge rejected fair use for Ross’s commercial use of Westlaw-derived editorial material to develop a competing, non-generative legal-research system. It is inaccurate to say the court ruled that all AI training on copyrighted works is unlawful.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Training is only one part of the copyright analysis
“AI copyright” often collapses several separate issues. Businesses should analyze them independently:
- Acquisition: How was the source material obtained? Was it licensed, public-domain, scraped, purchased, or supplied under contractual restrictions?
- Training or indexing: Was copying into the development process, search index, or retrieval system authorized or defensible under fair use?
- Output behavior: Does the system reproduce protected expression, or does it generate genuinely new material?
- Competition: Does the product substitute for the source owner’s product or threaten an existing licensing market?
A retrieval-augmented system creates its own issues involving document storage, indexing, access controls, display, attribution, and output generation. It should not automatically be treated as equivalent to pretraining a general-purpose model.
What the decision does not decide
- It does not establish that training every large language model on copyrighted material is infringement.
- It does not decide the copyright claims against every generative-AI company.
- It does not mean Westlaw owns court opinions or legal facts.
- It does not prove that every AI output resembling a copyrighted work is unlawful.
- It does not make an AI detector capable of determining whether a training use is fair use.
- It does not automatically resolve disputes involving licensed datasets, public-domain material, search engines, summaries, or retrieval systems.
Appeal and procedural status
On May 23, 2025, the district court certified interlocutory appellate questions involving the originality of Westlaw’s headnotes and Key Number System and Ross’s fair-use defense. The court stayed the case pending appellate proceedings while continuing to stand by its summary-judgment reasoning.
The last substantive court document reflected here is dated May 23, 2025. The case should not be described as fully over unless the current Third Circuit docket confirms a final disposition, such as acceptance and decision of the appeal, dismissal, or another resolution.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesRank #4
Even if the district-court ruling remains influential, it is a federal trial-court decision, not a nationwide rule that automatically binds every court or resolves every AI-training dispute.
Read the May 23, 2025 memorandum opinion.
Why the ruling matters to businesses
For publishers and copyright holders, the decision underscores the value of proprietary editorial additions—summaries, metadata, classification systems, annotations, and structured databases—not merely the underlying facts.
For AI companies, the risk is greatest when a product uses a competitor’s protected editorial work to offer a close substitute. Dataset provenance, licensing terms, access controls, and records showing how content was acquired become commercially important, not just legal housekeeping.
A licensed dataset can materially improve a company’s position, although a license does not automatically answer every question involving privacy, confidentiality, output reproduction, or downstream use.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Practical checklist for an AI or content business
- Identify which parts of the dataset are factual, public-domain, editorial, or otherwise protected.
- Separate underlying court opinions and facts from summaries, annotations, wording, and classification systems.
- Document how every important source was acquired.
- Review licenses, terms of service, access restrictions, and contractual limits.
- Assess whether the planned product competes with the source owner’s product or licensing market.
- Test whether the system can reproduce memorized passages or other protected expression.
- Keep provenance, permissions, filtering, and compliance records.
- Do not treat AI-detection or plagiarism scores as legal conclusions.
- Obtain advice from qualified copyright counsel before commercial deployment.
Tools and services: useful, but not legal answers
Businesses may use technology to investigate reuse, provenance, and policy violations, but these tools cannot determine whether an AI-training practice is fair use.
Best Value
Copyleaks
Copyleaks offers AI-text detection, plagiarism detection, AI-image detection, APIs, and enterprise workflows. Its pricing page listed Personal at $16.99 per month, or $13.99 per month when billed annually, and Pro at $99.99 per month, or $74.99 per month when billed annually; enterprise and education pricing is custom. The page stated that one credit covers up to 250 words or one image.
That may help publishers, agencies, schools, and enterprises identify material for further review. It cannot prove what was included in a model’s training corpus or decide a copyright claim.
Westlaw and Westlaw Precision
Thomson Reuters legal products are aimed at professional legal research, authoritative sources, editorial organization, citation tools, and related AI workflows. Pricing is generally quote-based and depends on users, jurisdictions, practice areas, and modules.
Lexis+ with Protégé
Lexis+ with Protégé is LexisNexis’s legal-AI offering for research, drafting, analysis, citation validation, and organization-specific documents. Lexis+ AI was renamed Lexis+ with Protégé in February 2026. The service directs prospective customers toward trials or contacting LexisNexis rather than publishing a standard public price.
Neither a premium legal-research platform nor a general-purpose AI assistant should be treated as a substitute for checking primary sources and obtaining legal advice on unsettled copyright questions.
The real lesson
Thomson Reuters’ victory is important because it shows how copyright risk can increase when a company takes a rival’s protected editorial compilation and uses it to build a commercial substitute.
It is much weaker evidence for the broader claim that all unauthorized generative-AI training is unlawful. The decisive facts here were the material involved, the non-generative system, the direct competitive relationship, and the court’s concern about market substitution.
For AI companies and content businesses, the prudent response is not to assume either automatic liability or automatic fair use. It is to examine the source material, acquisition method, product purpose, output behavior, competitive effect, and licensing record separately.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
