OpenAI publishers copyright lawsuit: legal battle

What happened

The U.S. Court of Appeals for the Third Circuit affirmed a ruling on September 29, 2026, that a technology firm infringed on Thomson Reuters' copyrights by using its proprietary legal summaries to train an artificial intelligence system 1. The appellate court found that the "fair-use shield" did not apply because the resulting AI product served the same commercial function as the original human-curated database 1. This decision upheld a lower court's finding that the developer's use of the content was not transformative, as it sought to replicate the utility of the Thomson Reuters platform rather than create a distinct new purpose 1. The ruling specifically addressed the "commercial job" performed by the data, concluding that when an AI model is built to perform the same task as the source material, it constitutes a market substitute rather than a protected innovation 1.

Why it matters

The Third Circuit's decision establishes a legal precedent that challenges the "transformative" fair-use defense frequently cited by AI developers in the Southern District of New York 1. By focusing on whether an AI product performs the same commercial function as the training data, the court provides a framework for publishers to argue that AI models are market substitutes that threaten the economic foundations of content suppliers 1. The ruling directly addresses the "human-reader" analogy, suggesting that while humans may learn from data to create new works, AI systems that automate the specific utility of a proprietary database do not enjoy the same legal latitude 1. For the parties in the consolidated SDNY litigation, this appellate victory for a content provider offers a counter-argument to the Department of Justice's September 1 assertion that AI training is a national security priority that should be protected as fair use 1. The decision also aligns with internal Microsoft research cited in the SDNY case, which found that AI-integrated search resulted in 90 percent fewer clicks to original news articles, reinforcing the argument that these systems act as commercial substitutes 1.

What's next

The parties in the consolidated Authors Guild action against OpenAI and Microsoft are scheduled to submit summary judgment reply briefs by November 6, 2026 1. These filings will likely address how the Third Circuit's "commercial function" test applies to the training of large language models on books and news articles 1. The District Court for the Southern District of New York is expected to issue rulings on the publishers' motion to disregard the DOJ's statement of interest and the AI Rights Project's motion for leave to file an amicus brief 1. A final decision on the cross-motions for summary judgment would settle whether the training stage of AI development is inherently infringing or if it requires a case-by-case analysis of the commercial impact on the original creators 1.

What happened

The AI Rights Project filed a motion for leave to submit an amicus brief in the In re OpenAI, Inc. Copyright Infringement Litigation on October 15, 2026 1. The filing, docketed as 2099 in the Southern District of New York, challenges the legal premises used in the Department of Justice’s September 1 statement of interest 1. While the government argued that training large language models is a national security priority that qualifies as fair use, the AI Rights Project asserts that national interests are not present in the fair use statute 1.

The proposed brief questions whether the transformative-use inquiry requires human creative judgment 1. It cites the 2025 Bartz v. Anthropic PBC decision, where a California court likened AI training to a "reader aspiring to be a writer," and argues that prior courts failed to examine if this human-reader analogy is legally sound 1. The filing also references Thaler v. Perlmutter to note that the Copyright Act requires a human author 1. Furthermore, the motion argues that under the Intellectual Property Clause of the U.S. Constitution, Congress rather than the judiciary holds the authority to recalibrate the balance between scientific progress and the exclusive rights of authors 1. The AI Rights Project states it takes no position on the final outcome of the fair use question or the specific policy contentions made by the DOJ 1.

Why it matters

The intervention introduces a constitutional challenge to the judicial expansion of fair use for machine training 1. By citing Eldred v. Ashcroft, the brief argues that deciding how to pursue Copyright Clause objectives is a matter for the legislature 1. This position contests the DOJ's September 1 assertion that the court has a "strong interest" in rejecting infringement claims to maintain American AI dominance 1.

The filing impacts the legal framework for the pending cross-motions for summary judgment by questioning the "human-reader" analogy used in the Bartz and Kadrey rulings, which OpenAI cited on September 8 to support its defense 1. If the court adopts the AI Rights Project’s reasoning, the fair use defense would require a determination on whether machine-led "statistical relationships" can legally mirror human creative processes 1. The AI Rights Project, which accepts no funding from major AI developers, is also seeking to apply this "Human Baseline Principle" in the In re Mosaic LLM Litigation currently pending in the Northern District of California 1.

What's next

The deadline for interested parties to file motions for leave to submit amicus briefs is October 16, 2026 1. The district court will subsequently rule on whether to grant the AI Rights Project leave to file the brief and will address the publishers' September 28 motion to disregard the DOJ's statement of interest. A ruling from the U.S. Court of Appeals for the Ninth Circuit regarding the Kadrey and Bartz decisions would clarify the weight of the precedents cited by OpenAI. The court has not yet set a date for the hearing on the consolidated cross-motions for summary judgment.

What happened

The U.S. Court of Appeals for the Third Circuit ruled on September 29, 2026, that training an artificial intelligence system on a competitor’s proprietary, human-curated database constitutes copyright infringement when the resulting product performs the same commercial function 1. The decision affirmed a lower court ruling that the "fair-use shield" does not protect developers who use copyrighted editorial content to build tools that compete directly with the data source 1. The court found that Westlaw’s attorney-written "headnotes"—summaries of legal points within judicial opinions—possess the "creative spark" required for copyright protection 1.

The appellate panel determined that the use of 2,243 individual headnotes by the defunct startup ROSS Intelligence was only "minimally transformative" 1. While the machine-learning processing used a different mechanism than the original database, the court found the ultimate commercial purpose—helping researchers locate legal authority—remained identical 1. This ruling followed a June 11, 2026, oral argument where judges questioned whether training for a competing commercial purpose could be considered transformative under the first factor of the fair-use test 1.

In the Southern District of New York, a new copyright infringement lawsuit, Sullivan et al v. OpenAI Foundation, was filed on September 16, 2026 2. This filing adds to the consolidated litigation where OpenAI and Microsoft are already defending against claims from The New York Times, the Authors Guild, and other publishers.

Why it matters

The Third Circuit decision establishes the first federal appellate precedent directly addressing whether AI training constitutes fair use under the Copyright Act 1. By rejecting the fair-use defense for models that serve the same commercial job as their training data, the court creates a specific category of legal exposure for developers using human-curated content 1. The ruling suggests that the degree of liability depends on the commercial function of the AI system relative to the data it consumes 1.

For the publishers in the SDNY litigation, this precedent provides a framework to argue that OpenAI’s models are not transformative because they substitute for the original reporting and editorial work of news organizations. The Third Circuit's comparison of editorial summaries to a "sculptor's choice" reinforces the copyrightability of curated datasets, even when those datasets are derived from public or uncopyrightable facts 1. This may impact the valuation of training data licenses, as the court found that bypassing a license to train on proprietary content—as ROSS did by hiring a third party to create memoranda based on Westlaw—does not insulate a developer from infringement claims 1.

What's next

The SDNY court is scheduled to rule on several pending matters, including the publishers' motion to disregard the DOJ’s statement of interest and the cross-motions for summary judgment filed by the Authors Guild and OpenAI. Interested parties have until October 16, 2026, to file motions for leave to submit amicus briefs in the consolidated case.

The Third Circuit's ruling may now be cited as persuasive authority in the SDNY proceedings, particularly regarding the "transformative" nature of AI training. OpenAI’s answer to the allegations in the wikiHow lawsuit was due by September 17, 2026, and the court will likely integrate the new Sullivan et al v. OpenAI Foundation case into the master docket, In Re: OpenAI Inc. Copyright Infringement. Further clarity on the fair-use defense will depend on whether the Kadrey and Bartz rulings, currently under appeal, align with or diverge from the Third Circuit’s September 29 decision.

What happened

The U.S. Department of Justice (DOJ) filed a 20-page statement of interest on September 1, 2026, marking the federal government's first formal intervention in copyright litigation regarding artificial intelligence training 6. The DOJ advised the U.S. District Court for the Southern District of New York that training large language models on copyrighted works is "highly transformative" and constitutes fair use under Section 107 of the Copyright Act 6. The government explicitly argued that the act of training models on copyrighted material "in and of itself, does not violate copyright laws" 6.

On September 28, 2026, The New York Times and other publishers filed a response asking the court to disregard the DOJ's position 1. The publishers argued the government's brief dismissed a U.S. Copyright Office study in a footnote and failed to account for the Office's views on fair use 1. Internal Microsoft documents unsealed in the litigation on September 4, 2026, revealed that Brent Hecht, Microsoft’s director of applied science, warned colleagues that AI development had started a "doom loop" that could "threaten the economic foundations of its essential suppliers" 2. Additional unsealed communications showed OpenAI policy director Jack Clark stating the company was "creating systems that substitute for the labor of the people that define the culture of society" 2.

In summary judgment motions filed September 8, 2026, OpenAI cited 2025 California federal rulings in Kadrey v. Meta and Bartz v. Anthropic to support its fair-use defense 5. The New York Times countered that these precedents do not apply because ChatGPT outputs serve as commercial substitutes for original journalism, a claim of market harm not present in the cited Meta or Anthropic cases 5.

Why it matters

The DOJ's intervention signals a federal policy shift that prioritizes AI development as a national security and economic imperative 6. The government argued that an adverse ruling against OpenAI would hand a competitive advantage to foreign rivals 6. For the publishing industry, the litigation serves as a "real-money test" of whether to license content or litigate 5. While publishers like News Corp signed a five-year deal worth up to $250 million in May 2024, a victory for The New York Times could force AI developers toward significantly richer licensing structures 5. Conversely, an OpenAI win on fair-use grounds would remove the primary leverage publishers currently hold in negotiations 5.

The unsealed internal warnings from Microsoft and OpenAI executives provide the plaintiffs with evidence to argue that the defendants were aware of the potential for "market substitution" 2. Microsoft research cited in the filings found that AI-integrated search resulted in 90 percent fewer clicks to news articles compared to traditional search engines 2. Another expert cited in the documents found that referral traffic from Google Search and Google Discover dropped by at least 30 percent following the introduction of AI overviews 2.

What's next

The district court is scheduled to rule on the publishers' motion to disregard the DOJ filing and the competing motions for summary judgment 1, 5. Interested parties have until October 16, 2026, to file motions for leave to submit amicus briefs, which are restricted to 6,000 words. The legal status of the fair-use defense remains unsettled, as the Kadrey and Bartz rulings relied upon by OpenAI are currently under appeal 5. A ruling from the district court on whether AI training causes specific competitive harm to the news industry would establish the template for all future licensing negotiations between AI labs and content providers 5.

What happened

Lawyers for The New York Times and a coalition of news organizations filed a motion on September 28, 2026, requesting that the U.S. District Court for the Southern District of New York disregard a statement of interest submitted by the Department of Justice 1. The publishers argue the government’s intervention, which supported OpenAI’s fair use defense, was filed at an inappropriately early stage of the litigation 1. In the filing, the plaintiffs claim the DOJ’s position reflects a bias toward artificial intelligence developers and fails to account for the expertise of the U.S. Copyright Office 1. The publishers noted that the DOJ dismissed a comprehensive study by the Copyright Office in a single footnote, suggesting the government's legal stance was influenced by extraneous considerations rather than established copyright law 1. This motion follows the DOJ’s September 1 filing, which characterized the training of AI models on copyrighted data as a national security priority and a transformative use of material 1.

Why it matters

The publishers’ challenge to the DOJ intervention creates a direct conflict between executive branch policy and the regulatory findings of the U.S. Copyright Office 1. By asserting that the DOJ ignored the Copyright Office’s views, the plaintiffs are attempting to isolate the government’s "national security" argument from the technical application of the fair use doctrine 1. The filing claims that requiring AI developers to pay for copyrighted content would not hinder national technology goals, countering the government's assertion that licensing requirements could slow American AI development relative to global competitors 1. For the news industry, the outcome of this motion determines whether the court will weigh the DOJ’s policy-driven defense of AI training as a material factor in the summary judgment phase 1. If the court accepts the publishers' argument, the legal focus may shift back to the specific commercial impact of AI outputs on original journalism, rather than the broader geopolitical implications of model training 1.

What's next

The district court will determine whether to strike or limit the weight of the DOJ’s statement of interest as it considers the pending cross-motions for summary judgment. Interested parties have until October 16, 2026, to file motions for leave to submit amicus briefs, which are restricted to 6,000 words 1. A ruling from the court on the publishers' September 28 motion would clarify the role of executive branch policy in the ongoing litigation. OpenAI is scheduled to respond to the allegations of bias and the procedural validity of the DOJ's filing in its upcoming briefs. The court's eventual ruling on the cross-motions for summary judgment will settle whether the training stage of AI development constitutes infringement or protected fair use under current U.S. law.

Internal OpenAI communications unsealed in SDNY litigation

What happened

The New York Times and co-plaintiffs filed an updated version of their summary judgment motion on September 17, 2026, revealing previously redacted internal communications from OpenAI 2. The unsealed documents include a statement from the head of ChatGPT describing the advancement of artificial intelligence models as an "existential threat" to traditional publishers 2. These admissions were filed in the Southern District of New York as part of the consolidated case In Re: OpenAI Inc. Copyright Infringement 2. The filing follows the September 8, 2026, cross-motions for summary judgment where the Authors Guild and novelists joined publishers in seeking a final ruling on whether training AI models on copyrighted works constitutes infringement 2.

Why it matters

The unsealed admissions shift the evidentiary basis for the publishers' claim that OpenAI’s commercial products are intended to substitute for original reporting 2. By documenting that OpenAI leadership recognized the "existential threat" posed to the news industry, plaintiffs aim to undermine OpenAI’s fair use defense, which requires the court to consider the effect of the use upon the potential market for the copyrighted work 2. The disclosure of these internal views complicates the U.S. Department of Justice’s September 1, 2026, intervention, which argued that AI training is a "transformative" national security priority 1. The new evidence supports the publishers' legal position that the technology firms did not merely innovate but created commercial substitutes that threaten the economic viability of journalism 2.

What's next

The district court is scheduled to review these unredacted admissions as it weighs the pending cross-motions for summary judgment 2. Outside parties seeking to influence the court’s interpretation of these internal documents have until October 16, 2026, to file motions for leave to submit amicus briefs 1. OpenAI is also required to address related copyright allegations in the wikiHow matter, with a response deadline of September 17, 2026, though the court has not yet confirmed the filing of that specific answer 2. A ruling on the summary judgment motions will determine if the case proceeds to trial or moves directly to an appellate court 1.

What happened — the development in detail

The Seattle Times Company and Newsday LLC initiated a new copyright infringement lawsuit against Microsoft and ten OpenAI entities in the Southern District of New York on September 4, 2026 1. The complaint names OpenAI Inc., OpenAI GP, LLC, OpenAI LLC, OpenAI OpCo, LLC, OpenAI Global LLC, OAI Corporation, OpenAI Holdings, LLC, OpenAI Foundation, and OpenAI Group PBC as defendants 1. The plaintiffs paid a filing fee of $405.00 to open the case, docketed as 1:26-cv-07644 1.

The initial filing encountered immediate procedural hurdles as the court clerk recorded multiple filing errors 1. On September 8, 2026, the court flagged the complaint for a deficient pleading and a signature error 1. The clerk also issued notices regarding deficient Rule 7.1 Corporate Disclosure Statements for both Newsday LLC and The Seattle Times Company 1. These notices stated that the attorney who signed the PDF was not the same attorney who electronically filed the documents, and that corporate parents or affiliates were not correctly added to the electronic record 1. The court also identified a PDF error in the plaintiffs' request for the issuance of summonses to the ten OpenAI entities and Microsoft 1.

Why it matters

The entry of The Seattle Times Company and Newsday LLC expands the coalition of regional and national publishers challenging the legality of AI training models 1. By filing in the Southern District of New York, these plaintiffs align their claims with the existing consolidated litigation, In Re: OpenAI Inc. Copyright Infringement, which already involves The New York Times and the Authors Guild. The procedural errors identified by the clerk on September 8, 2026, delayed the formal issuance of summonses, momentarily stalling the start of the 21-day window for OpenAI and Microsoft to file their responsive pleadings 1.

The inclusion of Newsday LLC and The Seattle Times Company suggests that regional news organizations are adopting the same legal strategy as larger outlets, arguing that the commercial use of their reporting to train generative models constitutes infringement rather than fair use. The court’s requirement for corrected corporate disclosure statements ensures transparency regarding the ownership structures of these media entities as the litigation moves toward potential consolidation or joint discovery phases 1.

What's next

The plaintiffs' attorney must re-file the Rule 7.1 Corporate Disclosure Statements and the request for issuance of summonses to correct the signature and PDF errors identified by the court 1. Once the summonses are properly issued and served, OpenAI and Microsoft will have a standard deadline to answer the complaint or move to dismiss. The court has not yet issued a scheduling order for this specific case, but it is expected to be coordinated with the ongoing summary judgment motions in the broader consolidated publisher litigation. The clerk's office continues to review the AO 121 Copyright and AO 120 Trademark forms submitted by the plaintiffs on September 4, 2026 1.

What happened

OpenAI, Microsoft, and a group of authors and news organizations filed competing motions for summary judgment on September 4, 2026 1. The filings occurred in the Southern District of New York multidistrict litigation In re OpenAI, Inc. Copyright Infringement Litigation 1.

The publishers and authors argued that OpenAI and Microsoft copied protected works without permission or payment to build commercial products 1. They contended that this unauthorized copying for model development is not fair use and damages existing and potential markets 1.

OpenAI and Microsoft argued that model training is highly transformative under copyright law 1. The technology companies stated that large language models learn statistical relationships and patterns rather than offering the original works as market substitutes 1.

Separately, the Center for AI and Digital Policy (CAIDP) announced it is preparing an amicus brief 1. The brief will address transparency, accountability, and protections for human creators 1. Judge Sidney H. Stein previously established an October 16, 2026 deadline for all motions for leave to file amicus briefs 1. The court limited these proposed briefs to 6,000 words 1.

The filings follow a U.S. Department of Justice Statement of Interest submitted on September 1, 2026 1. The government brief urged the court to reject any rule that treats model training on copyrighted text as automatic infringement 1. The government argued that broad fair use applications support innovation, competition, scientific progress, and national security 1.

What changed vs. the baseline

The litigation moved from the pleading stage to active summary judgment contentions. Previously, the baseline status rested on a November 2025 ruling that denied OpenAI's motion to dismiss direct infringement claims regarding AI outputs [Baseline]. The September 4, 2026 filings forced both sides to submit their core evidentiary records for a final judgment on liability 1.

The timeline for the Department of Justice intervention contains a minor date clarification. The baseline recorded the DOJ Statement of Interest on September 2, 2026 [Baseline]. Court records show the United States actually filed the statement on September 1, 2026 1.

The invitation for amicus briefs is not a new development. Judge Stein issued the order setting the October 16, 2026 deadline on August 7, 2026, nearly a month before the baseline report 1.

Why it matters

This case serves as the primary legal test of whether using copyrighted books and news articles to train generative AI systems qualifies as fair use under 17 U.S.C. § 107 1. The outcome will directly shape the economic rights of writers, the viability of licensing markets, and the operational rules for data acquisition in generative AI 1.

The litigation targets distinct acts across the entire AI lifecycle 1. The court must rule on the legality of acquiring works, constructing datasets, training models, distributing copies, and generating outputs that compete with human creators 1.

The DOJ intervention introduces federal industrial policy into a statutory copyright dispute 1. However, the government filing does not dictate the legal outcome 1. Judge Stein remains solely responsible for applying the Copyright Act to the specific evidence presented by the publishers and tech firms 1. The court is not required to adopt a categorical rule; it can rule that some training practices infringe while others constitute fair use 1.

Open questions

The court has not yet ruled on the cross-motions for summary judgment 1. It is unclear whether Judge Stein will decide the case on the current record or push the claims to a full trial.

The total number of outside groups participating in the case remains unconfirmed. While CAIDP confirmed its intent to file an amicus brief, other industry groups and creators have until the October 16, 2026 deadline to submit their requests 1.

The record does not show how the court will handle the publishers' July 2026 motion for sanctions against OpenAI [Baseline]. The impact of those alleged evidence obstruction claims on the pending summary judgment motions remains unknown [Baseline, 1].

What happened

On September 1, 2026, the U.S. Department of Justice (DOJ) filed a 20-page Statement of Interest in the consolidated OpenAI copyright litigation in the Southern District of New York 24. The filing, submitted under 28 U.S.C. § 517, addresses all 19 actions currently pending in the multidistrict litigation (MDL 25-md-3143), including the high-profile suit brought by The New York Times 2.

The DOJ brief argues that copying copyrighted text for the specific purpose of training a Large Language Model (LLM) constitutes "transformative" fair use 23. The government asserts that a ruling against OpenAI would threaten U.S. national security and global leadership in AI by creating prohibitive barriers to entry 26.

Key specifics of the filing and related events include:

• Stage Distinction: The DOJ divides AI development into three stages: acquisition, training, and output 2. The government explicitly limits its support of fair use to the training stage, leaving the legal status of the acquisition stage unaddressed 2.
• National Security Argument: The brief cites executive orders and the President’s National Policy Framework for Artificial Intelligence, arguing that rules making it "significantly more difficult" to develop AI in the U.S. jeopardize national interests 2.
• International Divergence: On September 2, 2026, one day after the filing, G20 innovation ministers in Chapel Hill adopted a consensus statement that notably omitted the phrase "training data," signaling a lack of international policy alignment with the U.S. litigation position 2.
• Publisher Response: The New York Times issued a statement on September 2, 2026, accusing the administration of favoring "trillion-dollar AI companies" over content creators 6.

What changed vs. the baseline

The primary change is the formalization of the DOJ's position across the entire consolidated docket. While the baseline noted the intervention, the latest details clarify that the government is isolating the "training" stage from "acquisition" 2.

• Not New: The core argument that training is fair use was already the prevailing defense for OpenAI and Microsoft 3. The November 2025 ruling allowing claims regarding AI outputs to proceed remains the standing judicial precedent for this case 5.
• New Distinction: The DOJ’s refusal to defend the "acquisition" stage is a new strategic pivot 2. This follows a precedent set by the Anthropic settlement approved by Judge William Alsup, where Anthropic was ordered to pay $1.5 billion (approximately $3,000 per work) for using "shadow libraries" to acquire data, even though the training itself was considered fair use 23.
• New Regulatory Context: The filing coincides with the August 2, 2026, implementation of EU disclosure requirements (Article 101), which carries fines of up to 3% of global turnover for failing to disclose training data provenance 2.

Why it matters

The DOJ’s intervention shifts the litigation's focus from whether training is legal to how the data was obtained. By defending training but staying silent on acquisition, the government has signaled that AI companies may still face massive liability if their data-gathering methods (such as using unauthorized "shadow libraries") are found to be illegal 23.

For stakeholders, this means "provenance"—the documented route of how data was acquired—is now the critical legal vulnerability 2. The $1.5 billion Anthropic settlement serves as a price discovery mechanism for this specific liability 2. Furthermore, the divergence between the U.S. DOJ's stance and the G20's omission of "training data" suggests that U.S. AI companies may face a fragmented global regulatory landscape where training is protected at home but penalized abroad 2.

Open questions

• Acquisition Liability: It remains unclear if Judge Sidney H. Stein will adopt the DOJ’s three-stage framework (acquisition, training, output) or if he will treat the process as a single infringing act 2.
• Sanctions: The court has yet to rule on the July 2026 request by publishers for sanctions against OpenAI regarding alleged evidence obstruction 8.
• International Compliance: With only 3 of 7 major AI providers currently providing prose answers to EU disclosure forms as of August 2026, it is unclear if OpenAI will face European fines while simultaneously being defended by the U.S. government 2.

What happened

OpenAI received formal service on August 27, 2026, in a new copyright infringement lawsuit filed by wikiHow, Inc. 3. The court confirmed service for nine separate OpenAI entities on September 8, 2026, including OpenAI, Inc., OpenAI LLC, and OpenAI OpCo, LLC 3. Service was accepted by agent Tatiyana Smith 3. OpenAI must file its answer to the wikiHow complaint by September 17, 2026 3.

On September 4, 2026, The New York Times Company, Daily News LP, and Ziff Davis Inc. filed redacted documents related to a motion for summary judgment in the consolidated proceedings 3. This followed a September 3, 2026, omnibus sealing order issued by Judge Sidney H. Stein to govern summary judgment and Daubert briefing 3.

What changed vs. the baseline

The litigation has expanded beyond news and media publishers to include instructional content platforms. The wikiHow filing on August 21, 2026, introduced a new plaintiff into the SDNY proceedings under cause 17:101 3.

Several items in the record are not new developments:

• Discovery Disputes: OpenAI’s demand for reporters’ notes and interview memos dates back to July 29, 2024 1. The Times’ opposition to these requests as "harassing" is a prior record 1.
• Oral Arguments: The order setting oral arguments for motions to dismiss and strike for October 8, 2025, is a historical event from the previous year 2.
• DOJ Intervention: The Department of Justice’s statement of interest supporting OpenAI was filed on September 2, 2026, and was already established in the baseline 36.

Why it matters

The entry of wikiHow indicates that OpenAI faces a broadening front of copyright claims. While news publishers focus on the "investment of enormous amount of time" in reporting, instructional platforms like wikiHow represent a different category of "how-to" data used for LLM training 13.

The September 3 sealing order and subsequent redacted filings signal that the parties are moving into the summary judgment phase 3. This stage will force the court to address the DOJ’s recent argument that LLM training is a "transformative" fair use essential to national security against the publishers' claims of "stolen" content 367. The outcome of the discovery dispute regarding reporter notes will set a precedent for whether AI defendants can probe the human authorship and "veracity" of the works they are accused of infringing 1.

Open questions

It is currently unconfirmed how OpenAI will respond to the wikiHow complaint by the September 17 deadline—specifically whether it will move to consolidate this case immediately with the existing multidistrict litigation (MDL) 3.

The court has yet to rule on the specific discovery motion regarding the production of reporters' notes 1. Watch for a judicial order on this matter, as it will determine the scope of evidence plaintiffs must provide to prove their works are "authored by humans" and not already in the public domain 1. Finally, the impact of the DOJ’s intervention on Judge Stein’s eventual summary judgment ruling remains the primary variable for the legal status of AI training data 36.

What happened

The New York Times filed a formal rebuttal in the Southern District of New York on September 4, 2026, directly challenging the U.S. Department of Justice’s (DOJ) intervention in its copyright lawsuit against OpenAI and Microsoft 1. The Times alleged that the tech companies "stole" journalism to build commercial products that "substitute" for original reporting and "undermine" the news industry 1.

This filing follows a September 2, 2026, Statement of Interest from the DOJ, which urged Judge Sidney H. Stein to side with the AI developers 1. The DOJ argued that restricting Large Language Model (LLM) development through a "misunderstanding of fair use doctrine" would damage American economic mobility and scientific progress 1.

The litigation, which began in 2023, now centers on two distinct phases of AI development:

• Input (Pre-training): The process where models ingest large-scale data to learn language patterns 1.
• Output (Post-training): The generation of human-like responses that the Times argues are often "substantially similar" to protected works 1.

OpenAI and Microsoft do not dispute using the publisher's content but maintain the practice is legally permissible under the fair use doctrine 1. The Times is seeking monetary damages, a permanent injunction against further training on its material, and the "destruction" of any models already trained on its work 1.

What changed vs. the baseline

New Developments:

• Formal Rebuttal: The New York Times’ September 4 filing is a new response to the DOJ’s September 2 intervention 1. It marks a shift from a bilateral dispute to a three-way legal argument involving the federal government.
• DOJ Policy Stance: While the baseline noted the DOJ's support for OpenAI, the full filing clarifies the government's specific fear: that a ruling for publishers would create a "barrier to entry" where only the largest firms could afford licensing, effectively stifling competition 1.

Prior Record (Not New):

• Attorney-Client Privilege Waiver: A ruling from November 2025 remains the standing precedent regarding evidence handling 2. The court previously found that OpenAI waived privilege concerning the deletion of training datasets derived from "shadow libraries" like Library Genesis (LibGen) 2.
• Willfulness Defense: The court’s 2025 decision that OpenAI cannot invoke privilege while simultaneously claiming "good faith" or "innocent infringement" remains the operative discovery standard 2.
• Crime-Fraud Exception: The court's 2025 denial of the "crime-fraud exception"—finding no probable cause that communications were made to further a crime—remains unchanged 2.

Why it matters

The DOJ’s intervention signals that the executive branch views AI training as a matter of national economic policy rather than a narrow copyright dispute 1. If the court adopts the DOJ’s view, it could "smooth the way" for universal AI training practices without the need for individual licenses 1.

Legal experts, including Harvard Law professor Lawrence Lessig, suggest that requiring rights clearance for all training data would be "disastrous" for innovation, as the current copyright system lacks an efficient way to identify and pay millions of individual owners 1. Conversely, a ruling in favor of the Times could force a massive "destruction" of existing models, potentially resetting the technical progress of OpenAI’s GPT series and Microsoft’s Copilot 1.

The case is one of over 100 active copyright lawsuits against AI companies, making the SDNY’s eventual ruling the primary bellwether for the entire creative economy 1.

Open questions

• Substantial Similarity: While Stanford Law professor Mark Lemley notes that AI outputs only "sometimes" copy original works, the court has yet to define the specific threshold for what constitutes an infringing output in a commercial chatbot 1.
• Evidence of Deletion: It remains unclear what the compelled discovery regarding OpenAI’s 2022 internal communications will reveal about the "non-use" or deletion of pirated datasets 2.
• Sanctions: The court has not yet ruled on the July 2026 motion by publishers seeking sanctions against OpenAI for alleged evidence obstruction.
• Legislative Interference: With the DOJ citing "American prosperity," it is unclear if Congress will attempt to codify AI fair use before Judge Stein issues a final summary judgment.