Skip to content
ContentLora

    Tip: press / anywhere to search.

    ● Developing story

    AI copyright lawsuits tracker: rulings, settlements and open cases

    Courts have so far split the question. Two US judges found that training on books was fair use on the records before them, while one sent pirated copies to trial and the other said the authors made the wrong arguments.[1][2][3] In September 2026 a federal appeals court, the Third Circuit, ruled on AI training. It held that ROSS Intelligence's use of Westlaw material to train a non-generative legal-research tool was not fair use.[4][5] Anthropic's $1.5 billion settlement with book authors received final approval on 20 July 2026, and first payments were expected by 15 November 2026.[6][7][8] The New York Times's 2023 suit alleges that OpenAI and Microsoft built their models by copying millions of its articles, and in the UK Getty Images lost its main copyright claim against Stability AI.[9][10]

    Editor reviewedStrict sourcingUpdated Developing25 claims · 92% primary14 entries · last entry Next check Nov 28, 2026Courts and the law, Artificial intelligence

    What we know

    • Judge Alsup held in June 2025 that training Anthropic's models on the plaintiffs' books was fair use[1]
    • The same order sent Anthropic's pirated library copies to trial[2]
    • The Bartz settlement is a non-reversionary $1.5 billion fund covering 482,460 works, about $3,000 per work[6]
    • The court granted final approval of the Bartz settlement on 20 July 2026[7]
    • Judge Chhabria ruled for Meta in June 2025 but said the authors made the wrong arguments[3]
    • On 29 September 2026 the Third Circuit held that ROSS Intelligence's use of Thomson Reuters's Westlaw materials to build a competing AI legal-research tool was not fair use[4]
    • The Third Circuit stressed that ROSS's AI was not generative and distinguished the models in Bartz and the OpenAI cases[5]
    • The UK High Court held that Stable Diffusion is not an infringing copy because it does not store or reproduce the works[10]
    • Universal Music Group and Udio announced in October 2025 that they had settled their litigation[20]
    • Warner Music Group said in November 2025 that its partnership with Suno settled their earlier litigation[21]

    What we don't know yet

    • When the court will rule on fair use in the New York Times case and the consolidated OpenAI cases in New York
    • Whether an appeals court will rule on fair use for training generative AI models; the Third Circuit's September 2026 Ross ruling concerned a non-generative tool
    • How the court in the consolidated OpenAI cases will treat the Justice Department's September 2026 statement of interest
    • Whether Getty Images will appeal the UK judgment
    • Whether the Bartz first payments went out by 15 November 2026 as expected
    • When the Copyright Office will publish the final version of Part 3 of its AI report

    As of 7 November 2026, the biggest AI copyright cases have produced two US fair-use rulings for AI companies, a $1.5 billion settlement, and one UK judgment against a rights holder.[1][11][6][10] In the flagship news case, The New York Times against OpenAI and Microsoft, the newspaper’s claims remain allegations; the court has narrowed one of them.[9][12] Since September 2026 there is also an appeals ruling: the Third Circuit held that training a non-generative legal-research tool on Westlaw material was not fair use.[4][5]

    This tracker follows major copyright cases over how AI models are trained and what they produce. It relies on court filings, judgments and official records. Claims made by parties are allegations unless a court has ruled on them, and this page is not legal advice.

    Books: Bartz v. Anthropic and Kadrey v. Meta

    In Bartz v. Anthropic, the court’s June 2025 order described anthropic as having downloaded millions of copyrighted books for free from pirate sites. It said the company had also bought and scanned print books to build a central library used for training.[13] On 23 June 2025, Judge William Alsup granted Anthropic summary judgment that the training use was fair use, as was the scanning of purchased books.[1] He denied summary judgment on the pirated library copies and set them for trial.[2]

    Two days later, Judge Vince Chhabria granted summary judgment to Meta in Kadrey v. Meta, a suit by 13 authors over training its Llama models.[11] He stressed that the ruling did not make Meta’s training lawful. In his words, it stood only for the proposition that these plaintiffs made the wrong arguments.[3]

    The $1.5 billion Anthropic settlement

    The Bartz case later settled. The settlement is a non-reversionary $1.5 billion fund covering 482,460 works, an estimated $3,000 or so per work.[6] On 20 July 2026, Judge Araceli Martínez-Olguín granted final approval and granted class counsel’s fee request in part.[7] The order counts 350 valid, timely opt-outs, covering 1,802 works.[14] The administrator expected to send first payments on or before 15 November 2026 for works without disputes among co-claimants.[8]

    News publishers: The New York Times v. OpenAI and Microsoft

    The New York Times Company sued microsoft and several openai entities in the US District Court for the Southern District of New York on 27 December 2023.[9] It alleges that their models were built by copying millions of its articles. It also alleges that ChatGPT and Copilot reproduce excerpts and summaries that cost it subscription, licensing, advertising and affiliate revenue.[9][15] According to a later OpenAI filing, in April 2025 the court dismissed one direct-infringement claim with prejudice only as it related to “abridgments”.[12] On 3 April 2025, multiple class actions against OpenAI were centralized in the same court as In re OpenAI, Inc., Copyright Infringement Litigation.[16] According to the Third Circuit, the US Justice Department filed a statement of interest in those consolidated cases on 1 September 2026. Relying on the Bartz ruling, it contended that training a large language model that can generate original responses is a transformative use.[17]

    Appeals: Thomson Reuters v. ROSS

    On 29 September 2026 the US Court of Appeals for the Third Circuit ruled on fair use and AI training in an interlocutory appeal. It held that Thomson Reuters’s Westlaw materials are copyrightable and that ROSS Intelligence’s use of them to build a competing AI legal-research platform was not fair use. It affirmed partial summary judgment for Thomson Reuters.[4] The court noted that ROSS’s AI was not generative: it returned passages from existing judicial opinions rather than creating new expression. It distinguished the models at issue in Bartz and the OpenAI cases, which can generate original expression.[5] The ruling therefore does not settle the question for generative AI models.

    Images and music

    In the UK, Getty Images sued Stability AI over Stable Diffusion. Getty acknowledged there was no evidence the model had been trained in the UK, so it dropped its training claim before judgment.[18] On 4 November 2025, the High Court dismissed Getty’s secondary copyright claim. It held that a model which does not store or reproduce the works is not an “infringing copy”.[10] Getty won in part on trade marks over generated watermarks, but the judge called those findings historic and extremely limited in scope.[19]

    Some record labels have settled with AI music services. Universal Music Group and Udio announced on 29 October 2025 that they had settled their litigation and would build a licensed platform.[20] Warner Music Group said on 25 November 2025 that its new partnership with Suno also settled their earlier litigation.[21]

    The US Copyright Office is publishing its AI report in parts. Part 1, on digital replicas, came out in July 2024. Part 2, on whether AI outputs can be copyrighted, followed in January 2025, and a pre-publication Part 3 on training in May 2025.[22] Part 3 says many training uses are likely transformative. It adds that commercial use of vast troves of works to make competing expressive content, especially through illegal access, goes beyond established fair use.[23]

    In the EU, providers of general-purpose AI models must instead put in place a policy to comply with EU copyright law and publish a summary of their training content, as the EU AI Act explainer sets out.[24] In the US, a December 2025 executive order directs a federal task force to challenge state AI laws, a separate fight followed in the US state AI laws tracker.[25]

    For the US statutes that regulate AI developers outside copyright, read the US state AI laws tracker next. The open-weight vs closed models explainer covers how model access shapes these disputes.

    Timeline

    14 confirmed

    1. confirmed

      Third Circuit rules training a non-generative AI tool on Westlaw material was not fair use[4][5]

    2. confirmed

      Justice Department files a statement of interest in the OpenAI cases[17]

      Known here from the Third Circuit's description in its Ross opinion.

    3. confirmed

      Court grants final approval of the $1.5bn Anthropic settlement[7][6][14]

    4. confirmed

      Warner Music Group and Suno settle[21]

    5. confirmed

      UK High Court rejects Getty's main copyright claim against Stability AI[10][19]

    6. confirmed

      Universal Music Group and Udio settle[20]

    7. confirmed

      Kadrey v. Meta, summary judgment for Meta on fair use[11][3]

    8. confirmed

      Bartz v. Anthropic, training held fair use; pirated copies sent to trial[1][2]

    9. confirmed

      Copyright Office releases pre-publication Part 3, on AI training[22][23]

    10. confirmed

      Court narrows one claim in the New York Times case[12]

      According to a later OpenAI filing, a direct-infringement claim was dismissed with prejudice only as it related to "abridgments".

    11. confirmed

      Class actions against OpenAI centralized in New York[16]

    12. confirmed

      Copyright Office publishes Part 2, on copyrightability of AI outputs[22]

    13. confirmed

      Copyright Office publishes Part 1 of its AI report, on digital replicas[22]

    14. confirmed

      The New York Times sues Microsoft and OpenAI[9][15]

      The complaint alleges the companies' models were built by copying millions of Times articles. The defendants have not been found liable.

    Next in the crash course

    More in Courts and the law

    Questions readers ask

    Is training AI on copyrighted books legal in the US?

    The rulings on books so far are from trial courts. In June 2025 two federal district judges ruled that the training uses before them were fair use. One of them, Judge Alsup, still sent Anthropic's pirated library copies to trial. The other, Judge Chhabria, said his ruling did not mean Meta's training was lawful, only that the authors made the wrong arguments. An appeals ruling from the Third Circuit in September 2026 went the other way for a non-generative legal-research tool trained on Westlaw material, and the court distinguished generative models like those in the Anthropic and OpenAI cases.[1][2][3][4][5]

    How much is the Anthropic settlement worth?

    The Bartz v. Anthropic settlement is a non-reversionary $1.5 billion fund covering 482,460 works. That works out to an estimated $3,000 or so per work. The court granted final approval on 20 July 2026.[6][7]

    When will authors in the Anthropic settlement be paid?

    The settlement administrator said it expected to send the first round of payments on or before 15 November 2026, for works where co-claimants do not disagree. Disputed works come in a later distribution.[8]

    What is the New York Times suing OpenAI and Microsoft for?

    The Times alleges that the companies built their models by copying millions of its articles, and that ChatGPT and Copilot reproduce excerpts and summaries that cost it subscription, licensing, advertising and affiliate revenue. These are allegations, not court findings.[9][15]

    Did Getty Images win its UK case against Stability AI?

    Mostly not. In November 2025 the High Court dismissed Getty's secondary copyright claim, holding that Stable Diffusion is not an "infringing copy". Getty won in part on trade marks, but the judge called those findings historic and extremely limited.[10][19]

    What does the US Copyright Office say about AI training?

    In its May 2025 pre-publication report, the Office said many training uses are likely transformative. It added that commercial use of vast troves of works to produce competing expressive content, especially through illegal access, goes beyond established fair use. Its reports are guidance, not court rulings.[23]

    Sources

    · 25 claims from 14 publishers

    Each numbered claim is a statement we checked against the sources listed with it. Tap a number in the text to see its sources in place.

    1. [1]

      On 23 June 2025, Judge William Alsup granted Anthropic summary judgment that using the plaintiffs' books to train its models was fair use, and that converting purchased print books to digital form was also fair use. confirmedas of 2025-06-23

    2. [2]

      The same order denied summary judgment on the pirated copies Anthropic kept in its central library and set them for trial. confirmedas of 2025-06-23

    3. [3]

      Judge Chhabria said the ruling did not mean Meta's training was lawful, only that these plaintiffs made the wrong arguments and failed to build a record for the right one. confirmedas of 2025-06-25

    4. [4]

      On 29 September 2026, in an interlocutory appeal in Thomson Reuters v. ROSS Intelligence (No. 25-2153), the US Court of Appeals for the Third Circuit held that Thomson Reuters's Westlaw materials are copyrightable and that ROSS's use of them to build a competing AI legal-research platform was not fair use, affirming partial summary judgment for Thomson Reuters. confirmedas of 2026-09-29

    5. [5]

      The Third Circuit noted that ROSS's AI was not generative AI, since it only returned passages from existing judicial opinions, and distinguished it from the AI models at issue in Bartz v. Anthropic and In re OpenAI, which can generate original expression. confirmedas of 2026-09-29

    6. [6]

      The Bartz v. Anthropic class settlement created a non-reversionary fund of $1.5 billion covering 482,460 works, an estimated payment of about $3,000 per work. confirmedas of 2026-07-20

    7. [7]

      On 20 July 2026, Judge Araceli Martínez-Olguín granted final approval of the Bartz v. Anthropic settlement and granted class counsel's fee motion in part. confirmedas of 2026-07-20

    8. [8]

      As of November 2026, the settlement administrator expected to send the first round of payments on or before 15 November 2026 for works without unresolved disputes among claimants. confirmedas of 2026-11-07

    9. [9]

      The New York Times Company sued Microsoft and several OpenAI entities in the US District Court for the Southern District of New York on 27 December 2023 (No. 1:23-cv-11195), alleging that their large language models were built by copying millions of its copyrighted articles. confirmedas of 2023-12-27

    10. [10]

      On 4 November 2025, Mrs Justice Joanna Smith dismissed Getty's secondary copyright infringement claim, holding that a model such as Stable Diffusion, which does not store or reproduce the copyright works, is not an "infringing copy". confirmedas of 2025-11-04

    11. [11]

      On 25 June 2025, Judge Vince Chhabria granted Meta summary judgment in Kadrey v. Meta, a suit by 13 authors, finding that Meta's use of their books to train its Llama models was fair use on the record before him. confirmedas of 2025-06-25

    12. [12]

      According to a later OpenAI filing, an opinion of 4 April 2025 in New York Times Co. v. Microsoft Corp. dismissed with prejudice one direct-infringement claim only insofar as it related to "abridgments". confirmedas of 2025-07-14

    13. [13]

      In Bartz v. Anthropic, the court's June 2025 summary-judgment order described Anthropic as having downloaded millions of copyrighted books for free from pirate sites and having also bought and scanned print books to build a central library, with some books used to train its language models. confirmedas of 2025-06-23

    14. [14]

      The final approval order records 350 valid, timely opt-outs, covering 1,802 works. confirmedas of 2026-07-20

    15. [15]

      The Times's complaint alleges that Microsoft's Copilot and OpenAI's ChatGPT reproduce verbatim excerpts and detailed summaries of its articles, depriving it of subscription, licensing, advertising and affiliate revenue. confirmedas of 2023-12-27

    16. [16]

      On 3 April 2025, multiple class actions against OpenAI were centralized as a multidistrict litigation in the Southern District of New York, In re OpenAI, Inc., Copyright Infringement Litigation (No. 25-md-3143). confirmedas of 2025-07-14

    17. [17]

      According to the Third Circuit's Ross opinion, the US Justice Department filed a statement of interest on 1 September 2026 in In re OpenAI, Inc., Copyright Infringement Litigation (S.D.N.Y.), relying on Bartz v. Anthropic to contend that training a large language model that can generate original responses is a transformative use. confirmedas of 2026-09-29

    18. [18]

      In Getty Images v Stability AI in the High Court of England and Wales, Getty acknowledged there was no evidence that Stable Diffusion was trained in the UK, so its training claim was abandoned before judgment, along with its main output and database-right claims. confirmedas of 2025-11-04

    19. [19]

      The judge found that Getty succeeded in part on trade mark infringement over watermarks generated by some older model versions, but called those findings historic and extremely limited in scope. confirmedas of 2025-11-04

    20. [20]

      On 29 October 2025, Universal Music Group and the AI music service Udio announced agreements under which they settled their copyright infringement litigation and agreed to work on a licensed music-creation platform. confirmedas of 2025-10-29

    21. [21]

      On 25 November 2025, Warner Music Group announced a partnership with the AI music service Suno that it said also settled previous litigation between the companies. confirmedas of 2025-11-25

    22. [22]

      The US Copyright Office is issuing its Copyright and Artificial Intelligence report in parts; Part 1 on digital replicas was published on 31 July 2024, Part 2 on the copyrightability of AI outputs on 29 January 2025, and a pre-publication version of Part 3 on generative AI training on 9 May 2025. confirmedas of 2026-11-07

    23. [23]

      In the Part 3 pre-publication report, the Copyright Office said that various AI training uses are likely transformative, but that commercial use of vast troves of copyrighted works to produce expressive content competing with them in existing markets, especially through illegal access, goes beyond established fair use boundaries. confirmedas of 2025-05-09

    24. [24]

      Providers of general-purpose AI models must keep technical documentation, give information to downstream providers, put in place a policy to comply with EU copyright law, and publish a sufficiently detailed summary of the content used for training. confirmedas of 2024-07-12

    25. [25]

      Executive Order 14365, Ensuring a National Policy Framework for Artificial Intelligence, signed on 11 December 2025, directed the Attorney General to set up within 30 days an AI Litigation Task Force whose sole job is to challenge state AI laws inconsistent with the order's policy. confirmedas of 2025-12-11

    Revision history (3)
    1. Re-checked the Bartz settlement payment timetable on the administrator site and looked for new rulings or settlements in the tracked cases; no change.
    2. Added the Third Circuit's September 2026 ruling in Thomson Reuters v. ROSS, an appeals ruling on AI training and fair use, and the Justice Department's statement of interest in the OpenAI cases. The administrator still expects first Bartz payments on or before 15 November 2026.
    3. Page created.

    Created Nov 7, 2026. Last reviewed by an editor on Nov 8, 2026. Next scheduled review: Nov 28, 2026.

    Cite this page

    "AI copyright lawsuits tracker: rulings, settlements and open cases." ContentLora, updated Nov 14, 2026. https://contentlora.com/events/ai-copyright-lawsuits-tracker

    Spotted an error? Suggest a correction or emailcorrections@contentlora.com.