Seattle Times lawsuit filings landed in the Southern District of New York on 4 September 2026, and they carry a second plaintiff: Newsday. Together the two newspapers accuse OpenAI, eight affiliated OpenAI entities and Microsoft Corporation of copying hundreds of thousands of their articles into the training data behind ChatGPT, Copilot and Bing Chat. The case is docketed as No. 1:26-cv-07644 and the plaintiffs have demanded a jury.
The Seattle Times lawsuit is worth reading rather than skimming because it is unusually specific. It names the datasets, it names the crawlers, it reproduces an 88-word passage a model gave back verbatim, and it puts real traffic numbers on the record. It also arrived on the same day both sides in the consolidated New York Times case moved for summary judgment, which means the legal question underneath it may be answered before this complaint is even briefed.
This article works through the document itself. It covers what the Seattle Times lawsuit alleges, who filed it and why that is awkward, the seven counts and what each one has to prove, the evidence attached to them, the traffic and revenue figures behind the harm claim, where the case sits in a crowded docket, what the defendants have said, and what any of it means if you publish content or build products on a large language model. Every figure below comes from the Seattle Times lawsuit itself, the docket, or named reporting, all read on 6 September 2026.
Table of contents
- What the Seattle Times Lawsuit Actually Alleges
- Who Filed the Seattle Times Lawsuit, and Why It Is Awkward
- The Seven Counts Inside the Seattle Times Lawsuit
- The Evidence the Seattle Times Lawsuit Puts on the Record
- The Traffic and Revenue Numbers Driving the Seattle Times Lawsuit
- Where the Seattle Times Lawsuit Sits in a Crowded Docket
- What OpenAI and Microsoft Say About the Seattle Times Lawsuit
- What the Seattle Times Lawsuit Means If You Publish or Build
- Frequently Asked Questions About the Seattle Times Lawsuit
- References
What the Seattle Times Lawsuit Actually Alleges
The core accusation is straightforward and the complaint states it in the first ten paragraphs, before any legal theory arrives.
The Seattle Times lawsuit in one line
The Seattle Times Company and Newsday LLC say the defendants obtained copies of their journalism without permission, used those copies to train and ground commercial generative AI systems, stripped the copyright information out along the way, and now sell products that substitute for the reporting they consumed.
How the copying is said to have happened
The Seattle Times lawsuit alleges the defendants spent years “methodically scraping” the plaintiffs’ websites with automated bots, bypassing paywalls and ignoring terms of service to obtain what it calls the plaintiffs’ crown jewels. Seattle Times readers get one free pageview before payment is required. Newsday has run a complete hard gate since August 2022, under which no Newsday-produced editorial text is available to non-subscribers at all.
The datasets the Seattle Times lawsuit names on the record
Rather than gesturing at “the internet”, the filing names specific corpora: WebText, WebText2, Common Crawl, and Microsoft’s own Bing search index. It then alleges the defendants deliberately removed or altered copyright management information — article titles, author names and copyright notices — from those copies.
The framing the plaintiffs chose
Paragraph three sets the tone: “AI products like ChatGPT and CoPilot are touted as producers of content, but in fact they are rapacious consumers.” Paragraph four supplies the line every outlet quoted: “Like a snake eating its own tail, GenAI that is trained on painstakingly researched, expensive-to-produce content threatens to destroy the very news organizations by competing directly with them.”
| Element | As recorded on the docket |
|---|---|
| Case name | The Seattle Times Company v. OpenAI Inc. |
| Docket number | 1:26-cv-07644 |
| Court | U.S. District Court, S.D.N.Y. |
| Filed | 4 September 2026 |
| Plaintiffs | The Seattle Times Company, Newsday LLC |
| Defendants | Nine OpenAI entities plus Microsoft Corporation |
| Counsel | Klaris Law PLLC |
| Cause | 17:101 copyright infringement |
| Counts pleaded | Seven |
| Length | 38 pages, four exhibits |
| Jury demanded | Yes |
Who Filed the Seattle Times Lawsuit, and Why It Is Awkward
Two regional newspapers of very similar weight brought the Seattle Times lawsuit together, and one of them is suing an organisation that helps pay for its journalism.
The Seattle Times Company
Founded on 3 May 1886 and bought by Alden J. Blethen in August 1896, the paper has stayed in Blethen family hands for five generations. It has won 11 Pulitzer Prizes since 1950, reaches roughly 1.7 million adults a month across the Seattle-Tacoma market, and runs about 108,000 digital-only subscribers alongside roughly 100,000 print subscriptions and single-issue sales.
Newsday LLC
Newsday has served Long Island and New York City since 1940 from offices in Melville, New York. It holds 19 Pulitzer Prizes, sits among the ten largest daily-circulation US newspapers, and reports over 84,000 Sunday print copies, over 65,000 daily copies and over 60,000 digital subscribers. Its website records 51 million average monthly page views against 2.1 million monthly unique visitors — a very high pages-per-visitor ratio for a regional title.
The Seattle Times lawsuit names one of its own funders
Microsoft Philanthropies underwrites some Seattle Times journalism projects. In 2024, Microsoft and OpenAI jointly funded a $10 million Lenfest Institute AI fellowship whose inaugural participating newsrooms included both plaintiffs. The Times says it maintains editorial independence. Seattle Times Co. President and CEO Alan Fisco, who took the role on 1 January, told staff in a memo: “This was not an easy decision. However, we feel strongly that we must defend our content — which we spend millions of dollars a year to produce — from being used without our consent or compensation.”
The union’s reply
The Seattle Times Union, representing more than 160 employees, backed the filing and immediately used it. It said that in ongoing contract talks the company has refused to guarantee it will not replace non-reporter newsroom jobs with AI, adding: “If the Seattle Times Co. truly cares about the threat AI poses to journalism’s business model, it should protect the workers who produce the copyrighted material at the heart of this case.”
| Measure | The Seattle Times | Newsday |
|---|---|---|
| Founded | 1886 | 1940 |
| Pulitzer Prizes | 11 | 19 |
| Monthly page views | 29 million | 51 million |
| Monthly unique visitors | 6 million | 2.1 million |
| Digital subscribers | ~108,000 | 60,000+ |
| Paywall model | One free pageview | Hard gate since Aug 2022 |
| Trademark registered | 30 April 2019 | 25 March 1997 |
| Mark incontestable | Yes | Yes |
The Seven Counts Inside the Seattle Times Lawsuit
Most coverage described the Seattle Times lawsuit as a copyright case. It is broader than that, and the extra counts are where the interesting exposure sits.
Counts I and II: copyright, direct and vicarious
Count I pleads direct infringement under 17 U.S.C. §501 across reproduction for model training, fine-tuning, grounding, and the dissemination of output containing copies and derivatives. Count II names Microsoft and several OpenAI parent entities as vicariously liable, on the theory that Microsoft supplied and controlled the computing platform used to store and reproduce the corpus, then folded the resulting models into Copilot and Bing Chat.
Counts III and IV: the DMCA claims
These are the quieter and potentially costlier ones. Count III pleads removal of copyright management information under §1202(b)(1); Count IV pleads distribution of works knowing that information had been removed, under §1202(b)(3). DMCA statutory damages run per violation, and the Seattle Times lawsuit alleges hundreds of thousands of articles.
Counts V, VI and VII: trademark dilution
Count V pleads federal dilution under 15 U.S.C. §1125(c). Count VI pleads Washington state dilution under RCW 19.77.160 for the Seattle Times alone. Count VII pleads New York dilution under Gen. Bus. L. §360-L for Newsday alone. All three rest on the same factual hook: hallucinated output that misattributes fabricated content to the mastheads, which the plaintiffs call tarnishment.
What the Seattle Times lawsuit asks the court to do
Beyond damages and profits, the prayer for relief asks for a permanent injunction and — under 17 U.S.C. §503 — the impoundment or destruction of all copies of the works and of every model and training dataset that incorporates them. It also asks for treble damages and fees on the dilution counts on a bad-faith theory.
| Count | Statute | Turns on |
|---|---|---|
| I. Copyright infringement | 17 U.S.C. §501 | Whether training copies are fair use |
| II. Vicarious infringement | 17 U.S.C. §501 | Microsoft’s control of the infrastructure |
| III. CMI removal | 17 U.S.C. §1202(b)(1) | Knowing removal of author and notice data |
| IV. CMI distribution | 17 U.S.C. §1202(b)(3) | Distributing works with that data stripped |
| V. Federal dilution | 15 U.S.C. §1125(c) | Fame of the marks and hallucinated output |
| VI. Washington dilution | RCW 19.77.160 | Fame in Washington state |
| VII. New York dilution | NY Gen. Bus. L. §360-L | Secondary meaning in New York |
The Evidence the Seattle Times Lawsuit Puts on the Record
A complaint is not proof, but the Seattle Times lawsuit does more than assert. It attaches exhibits and describes reproducible tests.
The 88-word Boeing 737 MAX passage
The headline exhibit is a passage from the Seattle Times’ Pulitzer-winning coverage of the 2019 Boeing 737 MAX crashes. The complaint says a defendant model returned 88 consecutive words verbatim when prompted with nothing more than the article’s headline and its web address, and highlights the matching text. The plaintiffs draw the obvious conclusion: a reader determined to get past the paywall can simply ask.
Memorisation as the theory of copying
The filing argues that reproduction on demand proves a retrievable copy persists in the model’s parameters long after training, and that this defeats the characterisation of parameters as abstract statistical relationships. It reframes a natural language processing system as a container for copies rather than a set of abstractions, and it notes the test was built so the reproduced text could not have come from anything the user pasted in.
Retrieval as present-tense copying
Separately, the complaint alleges continuing infringement through retrieval-augmented generation: broad web crawls that build fresh indices, plus live scrapes of the plaintiffs’ sites at query time, neither of which preserves copyright information. This matters because it survives any argument that the training happened years ago.
The crawler and robots.txt record
The filing walks through OpenAI’s three published crawlers — GPTBot for training material, OAI-SearchBot for surfacing content in ChatGPT search, and ChatGPT-User for live fetches — and notes that Newsday’s own robots.txt instructs both OpenAI and Common Crawl not to crawl its site. The plaintiffs are building a paper trail of refusal.
Hallucination as brand damage
The last evidentiary strand is the one that powers the dilution counts. When a product invents content and attributes it to a masthead, or attributes a masthead’s real reporting to someone else, the Seattle Times lawsuit says the injury is not lost traffic but a trusted brand associated with material it never reviewed and would not publish.
The Traffic and Revenue Numbers Driving the Seattle Times Lawsuit
Fair use turns partly on market harm, so the Seattle Times lawsuit puts numbers where most complaints put adjectives.
The 47% referral collapse
Citing industry data for December 2024 to December 2025, the complaint says search referral traffic to mid-sized regional and metro daily publishers fell by approximately 47%, against approximately 22% for larger national publishers. The explanation offered is structural: mid-sized titles leaned harder on incidental search traffic, which AI-generated answers now keep.
A separate Search Engine Land dataset cited in the filing puts small-publisher search traffic down 60% over two years to March 2026 — a longer window than the other two figures, and a useful reminder that these bars are not measured over identical periods.
The subscription channel, and why it compounds
The second harm theory is conversion. Both papers turn readers into paying subscribers, and the Seattle Times lawsuit argues a reader who gets a satisfactory answer from a chatbot has less reason ever to subscribe — suppressing new growth and accelerating churn at the same time. The subscriber bases at stake are not large, which is exactly the point.
The licensing market they were left out of
The third argument is the strongest one procedurally. OpenAI has signed licensing deals with more than a dozen news and media organisations — the complaint names the Associated Press, News Corp, Axios, Axel Springer, The Atlantic, the Financial Times, Dotdash Meredith and Vox Media — and the publicly disclosed terms of just three of them exceed $300 million combined. A functioning licence market is hard to square with a defence that no licence was needed. Our earlier explainer on whether it is legal to train AI models on copyrighted books sets out the same tension in the book market.
Where the Seattle Times Lawsuit Sits in a Crowded Docket
This is not a first mover. It is a late entrant into a fight whose decisive ruling may already be written.
The consolidated cases before Judge Stein
The New York Times sued Microsoft and OpenAI on 27 December 2023. Raw Story Media, The Intercept and Daily News followed in 2024, along with the Center for Investigative Reporting. Ziff Davis filed in May 2025 and U.S. News & World Report in November 2025. All are consolidated as In re OpenAI, Inc. Copyright Infringement Litigation, No. 25-md-03143, before U.S. District Judge Sidney H. Stein in Manhattan.
The timing is the story
On the same Friday the Seattle Times lawsuit was filed, the publishers in the consolidated case and both defendants moved for summary judgment. A ruling on whether training on copyrighted journalism is fair use could therefore land within weeks, in a case these plaintiffs are not party to but whose reasoning will govern theirs.
The government picked a side first
Days earlier, the US Justice Department filed a brief siding with Microsoft and OpenAI, arguing that a ruling for the publishers would stifle American AI development. We covered that filing in detail when the government sided with OpenAI on fair use, and it is the single most important piece of context for reading the odds on the Seattle Times lawsuit.
The wider pattern beyond OpenAI
The same argument is being run against other model builders. CNN sued Perplexity for copyright infringement earlier this year, and the pattern extends past text — we looked at it in the video and streaming context when Twitch and Amazon were sued over streamers’ content. Meanwhile the Associated Press and Vox Media took the licensing route instead.
What OpenAI and Microsoft Say About the Seattle Times Lawsuit
Only one defendant has commented publicly so far, and it did so carefully.
What Microsoft said about the Seattle Times lawsuit
A Microsoft spokesperson said on the Friday evening: “While we’re surprised by the lawsuit, we appreciate the importance of the Seattle Times to our region and we’re always happy to sit down and explore solutions to this type of dispute.” That is an invitation to negotiate, not a defence, and the reference to the region acknowledges the funding relationship without addressing it.
OpenAI has not responded on the record
OpenAI had not issued a statement on the Seattle Times lawsuit at the time of writing. Its position in the consolidated litigation is well established: training is transformative, the outputs serve a different purpose from the originals, and copyright does not reach the statistical patterns a model learns.
The transformation question
The Seattle Times lawsuit attacks that directly in paragraph 14, arguing there is nothing transformative about copying journalism to build products that compete for the same readers, advertisers and subscription dollars. Whether repackaging counts as transformation is the question summary judgment in the New York Times case is expected to answer.
The substitution question
The second half of fair use is market effect, and this is where the 47% figure does its work. If AI answers demonstrably replace the click, the fourth fair use factor tilts toward the publishers regardless of how novel the technology is. If they do not, the whole harm theory thins out.
It is not clear anyone tried to negotiate
Reporting notes it is unclear whether licensing talks preceded the suit. Given that both plaintiffs sat in a fellowship programme funded by both defendants in 2024, the absence of a licence offer — or the rejection of one — is likely to surface in discovery.
What the Seattle Times Lawsuit Means If You Publish or Build
Most readers are neither a newspaper nor a frontier lab. The practical takeaways still land.
If you publish content, your technical record now matters
The complaint leans on Newsday’s robots.txt entries, the Seattle Times’ terms of service section 35, and the copyright notice on every page. None of those stopped anything, but all of them are now evidence of refusal. Publishing clear terms and crawler directives costs nothing and is the cheapest exhibit you will ever create.
Treat robots.txt as a signal, not a control
The filing itself describes the Robots Exclusion Protocol as a decades-old voluntary convention. It is a statement of intent that helps establish knowledge, not an access control. If content genuinely must not be copied, it belongs behind authentication, not behind a text file.
If you build on these models, read the indemnities
An injunction that reaches model weights and the corpus behind them is a live request in the Seattle Times lawsuit. It is unlikely to be granted in that form, but any organisation whose product depends on a third-party model should know what its vendor contract says about infringement claims, model withdrawal and continuity. That is a due-diligence question, not a legal one, and it belongs in your AI strategy work rather than in your legal budget.
Licensing is now a documented market with a price
The $300 million figure for three disclosed deals is the most quotable number in the Seattle Times lawsuit, because it converts an abstract question into a commercial one. Once a market rate exists, “we could not have licensed it” stops being available as an argument for anyone.
Watch the summary judgment ruling, not this docket
Nothing will move quickly here. The case that matters in the next few weeks is the consolidated one before Judge Stein. Whatever that ruling says about transformation and substitution will set the terms on which this complaint, and the next dozen like it, are read.
Frequently Asked Questions About the Seattle Times Lawsuit
Who exactly filed the Seattle Times lawsuit?
The Seattle Times Company and Newsday LLC filed jointly on 4 September 2026 in the Southern District of New York, represented by Klaris Law PLLC, against nine OpenAI entities and Microsoft Corporation.
How is the Seattle Times lawsuit different from the New York Times case?
The legal theory overlaps heavily, but this complaint adds three trademark dilution counts built on hallucinated attribution, and pleads regional publisher economics — the 47% referral decline — rather than national ones. It is also filed after the New York Times case reached summary judgment.
Are the plaintiffs asking for models to be destroyed?
Yes, as one form of relief. The prayer asks for impoundment or destruction under 17 U.S.C. §503 of all copies of the works and of every model and training dataset incorporating them, alongside damages and an injunction.
What is the strongest part of the Seattle Times lawsuit?
Probably the DMCA counts and the licensing-market evidence. Statutory damages for copyright management information removal accrue per violation, and OpenAI’s own signed deals make it hard to argue no licence market existed.
What is the weakest part?
The scale of relief and the dilution counts. Courts rarely order model destruction, and dilution claims built on hallucinated output are novel enough that they may not survive a motion to dismiss.
When will any of this be decided?
Not soon. A response to the complaint will take months, and discovery longer. The nearer-term signal is the summary judgment ruling expected in the consolidated case before Judge Sidney H. Stein.
References
CourtListener docket: The Seattle Times Company v. OpenAI Inc.
Seattle Times sues Microsoft and OpenAI, alleging they trained their AI on its journalism
Seattle Times and Newsday are the latest publications to sue OpenAI and Microsoft
Two more news organizations sue OpenAI and Microsoft for copyright infringement
Google Users Are Less Likely to Click on Links When an AI Summary Appears in the Results
Small Publisher Search Traffic Fell 60% Over Two Years: Data
Microsoft Bing Hits 100 Million Active Users in Bid to Grab Share from Google
More AI coverage: explore Progressive Robot's AI Models, Tools & Releases hub — hands-on reviews, setup guides and benchmarks in one place.