Loading
Technology
10 min read
Areas of Expertise :
Artificial Intelligence, Machine Learning, Copyright, Digital Forensics, Linguistics
Published Date:
Last Modified Date:
Jurisdiction:
Federal. Second Circuit
The action was filed on September 19, 2023, in the Southern District of New York by the Authors Guild and a group of individual authors who allege that OpenAI copied copyrighted books without permission and used those works in connection with training GPT models. On December 4, 2023, the plaintiffs amended their complaint to add Microsoft, alleging that Microsoft was involved in the deployment and commercialization of the AI systems at issue.
At its core, the case asks how copyright law applies when books are allegedly copied at scale to build and operate large language models. The plaintiffs contend that protected literary works were reproduced without authorization and then used in the development and distribution of commercial AI products. Their theory is not limited to training alone. It also reaches alleged infringing outputs and the broader economic effect that AI systems may have on authors and book-licensing markets.
The case later became part of the broader federal multidistrict litigation in New York involving copyright claims against OpenAI and, in some actions, Microsoft. That transfer did not resolve any merits issue, but it moved the Authors Guild action into a coordinated proceeding before Judge Sidney H. Stein. As of late March 2026, the litigation remains active, with no public final judgment, settlement, or dismissal ending the case.
What makes this matter especially important is that it places abstract policy debates into a live federal record. The dispute is no longer just about whether AI training raises copyright concerns in theory. It now involves concrete pleadings, motion practice, discovery disputes, and court rulings that may influence how future courts assess copying, outputs, market harm, and the use of copyrighted books in generative AI systems.
The case began when the Authors Guild and several individual authors filed a putative class action against OpenAI in the Southern District of New York. The complaint alleged that OpenAI copied copyrighted books without authorization and used those works in connection with training its GPT models. From the start, the suit framed the dispute as a direct challenge to the use of literary works in generative AI development rather than a broader complaint about technology policy.
The filing came from both a major writers’ organization and well-known individual authors. That combination gave the case immediate visibility and positioned it as a test case over whether authors can pursue copyright claims on a class-wide basis in the AI context. It also signaled that plaintiffs would try to connect technical model-development issues to traditional copyright doctrines familiar to federal courts.
The litigation expanded when the plaintiffs filed an amended complaint adding Microsoft as a defendant. The revised pleading alleged that Microsoft was not merely a remote business partner, but an active participant in the deployment and monetization of the AI systems identified in the lawsuit. That change widened the case beyond OpenAI’s alleged copying and placed more attention on downstream commercial use.
This amendment was important for another reason. It reflected a broader plaintiff strategy seen in AI copyright litigation: not only challenging the alleged ingestion of copyrighted works, but also targeting the commercial ecosystem built around the models. By bringing Microsoft directly into the pleading, the plaintiffs tied copyright arguments to product rollout, distribution, and monetization.
During 2024, the Authors Guild case moved forward as other copyright suits involving authors, publishers, and media entities were filed against OpenAI and Microsoft in federal court. Even before formal MDL transfer, related author cases in New York were already being coordinated for pretrial purposes. That procedural overlap showed that the litigation was growing in both scale and complexity.
The Authors Guild matter was no longer a standalone lawsuit. It became part of a growing body of AI copyright cases raising overlapping questions about copying, fair use, outputs, class treatment, and harm to licensing markets. For lawyers following the case, that meant strategy would increasingly be influenced by coordination across related actions rather than by one complaint alone.
A major procedural development occurred on April 3, 2025, when the Judicial Panel on Multidistrict Litigation transferred multiple copyright actions against OpenAI and Microsoft to the Southern District of New York. The Authors Guild action was included in that order and became part of In re OpenAI, Inc. Copyright Infringement Litigation, MDL No. 3143. The transfer did not decide any copyright issue on the merits, but it centralized overlapping factual and legal disputes before one court.
That change matters because MDL treatment does more than reorganize the docket. It alters litigation tempo, motion practice, and discovery planning. Instead of multiple courts handling similar disputes on parallel tracks, one judge now oversees consolidated pleadings and shared pretrial issues. In a case involving common questions about datasets, model development, corporate relationships, and alleged outputs, that coordination can materially affect leverage on both sides.
After centralization, the class plaintiffs filed a Consolidated Class Action Complaint on June 13, 2025. That pleading became a central document in the MDL because it combined overlapping class allegations into a single operative complaint for motion practice. It marked the shift from separate complaints toward a more unified framework for dismissal arguments, discovery disputes, and later certification issues.
In MDL practice, consolidation often forces plaintiffs to present their strongest, most coherent version of the case. It likewise gives defendants a clearer target for dismissal motions. Here, that process helped define which copyright theories would face early judicial testing.
One of the most closely watched rulings in the MDL came on October 27, 2025, when Judge Stein denied OpenAI’s motion to dismiss the consolidated class plaintiffs’ output-based direct copyright infringement claim. The court did not find liability, and it did not resolve the broader legality of AI training on copyrighted books. But it did allow that specific theory to move past the pleading stage.
The ruling focuses on whether the AI system’s generated responses can themselves support a direct infringement theory tied to protected works. Allowing that theory to proceed does not establish that the plaintiffs will ultimately prove copying, substantial similarity, causation, or recoverable harm. It does mean, however, that the court found the allegations sufficient at the pleading stage to justify further litigation on that issue.
Public filings in late 2025 and early 2026 show that the litigation has continued beyond the motion-to-dismiss stage and into active discovery disputes. Among the public issues addressed in the MDL were disputes concerning deleted datasets, privilege assertions, and the scope of discovery into how certain book-related datasets were handled. Those disputes underscore that this litigation will likely turn not only on legal doctrine, but also on technical records, internal communications, and evidentiary detail.
As of late March 2026, there has been no public final judgment, settlement, or dismissal ending the Authors Guild matter or the broader MDL. The case remains active. At this point, the dispute stands as an ongoing federal test of how copyright law may apply to books, AI model development, generated outputs, and commercial deployment in the generative AI setting.
Court:
United States District Court for the Southern District of New York
Case Name:
Authors Guild et al. v. OpenAI Inc. et al.
Copyright Infringement
Publishing
Licensing
Plagiarism
Computer Linguistics
Data Mining Algorithm
The action was filed on September 19, 2023, in the Southern District of New York by the Authors Guild and a group of individual authors who allege that OpenAI copied copyrighted books without permission and used those works in connection with training GPT models. On December 4, 2023, the plaintiffs amended their complaint to add Microsoft, alleging that Microsoft was involved in the deployment and commercialization of the AI systems at issue.
At its core, the case asks how copyright law applies when books are allegedly copied at scale to build and operate large language models. The plaintiffs contend that protected literary works were reproduced without authorization and then used in the development and distribution of commercial AI products. Their theory is not limited to training alone. It also reaches alleged infringing outputs and the broader economic effect that AI systems may have on authors and book-licensing markets.
The case later became part of the broader federal multidistrict litigation in New York involving copyright claims against OpenAI and, in some actions, Microsoft. That transfer did not resolve any merits issue, but it moved the Authors Guild action into a coordinated proceeding before Judge Sidney H. Stein. As of late March 2026, the litigation remains active, with no public final judgment, settlement, or dismissal ending the case.
What makes this matter especially important is that it places abstract policy debates into a live federal record. The dispute is no longer just about whether AI training raises copyright concerns in theory. It now involves concrete pleadings, motion practice, discovery disputes, and court rulings that may influence how future courts assess copying, outputs, market harm, and the use of copyrighted books in generative AI systems.
The case began when the Authors Guild and several individual authors filed a putative class action against OpenAI in the Southern District of New York. The complaint alleged that OpenAI copied copyrighted books without authorization and used those works in connection with training its GPT models. From the start, the suit framed the dispute as a direct challenge to the use of literary works in generative AI development rather than a broader complaint about technology policy.
The filing came from both a major writers’ organization and well-known individual authors. That combination gave the case immediate visibility and positioned it as a test case over whether authors can pursue copyright claims on a class-wide basis in the AI context. It also signaled that plaintiffs would try to connect technical model-development issues to traditional copyright doctrines familiar to federal courts.
The litigation expanded when the plaintiffs filed an amended complaint adding Microsoft as a defendant. The revised pleading alleged that Microsoft was not merely a remote business partner, but an active participant in the deployment and monetization of the AI systems identified in the lawsuit. That change widened the case beyond OpenAI’s alleged copying and placed more attention on downstream commercial use.
This amendment was important for another reason. It reflected a broader plaintiff strategy seen in AI copyright litigation: not only challenging the alleged ingestion of copyrighted works, but also targeting the commercial ecosystem built around the models. By bringing Microsoft directly into the pleading, the plaintiffs tied copyright arguments to product rollout, distribution, and monetization.
During 2024, the Authors Guild case moved forward as other copyright suits involving authors, publishers, and media entities were filed against OpenAI and Microsoft in federal court. Even before formal MDL transfer, related author cases in New York were already being coordinated for pretrial purposes. That procedural overlap showed that the litigation was growing in both scale and complexity.
The Authors Guild matter was no longer a standalone lawsuit. It became part of a growing body of AI copyright cases raising overlapping questions about copying, fair use, outputs, class treatment, and harm to licensing markets. For lawyers following the case, that meant strategy would increasingly be influenced by coordination across related actions rather than by one complaint alone.
A major procedural development occurred on April 3, 2025, when the Judicial Panel on Multidistrict Litigation transferred multiple copyright actions against OpenAI and Microsoft to the Southern District of New York. The Authors Guild action was included in that order and became part of In re OpenAI, Inc. Copyright Infringement Litigation, MDL No. 3143. The transfer did not decide any copyright issue on the merits, but it centralized overlapping factual and legal disputes before one court.
That change matters because MDL treatment does more than reorganize the docket. It alters litigation tempo, motion practice, and discovery planning. Instead of multiple courts handling similar disputes on parallel tracks, one judge now oversees consolidated pleadings and shared pretrial issues. In a case involving common questions about datasets, model development, corporate relationships, and alleged outputs, that coordination can materially affect leverage on both sides.
After centralization, the class plaintiffs filed a Consolidated Class Action Complaint on June 13, 2025. That pleading became a central document in the MDL because it combined overlapping class allegations into a single operative complaint for motion practice. It marked the shift from separate complaints toward a more unified framework for dismissal arguments, discovery disputes, and later certification issues.
In MDL practice, consolidation often forces plaintiffs to present their strongest, most coherent version of the case. It likewise gives defendants a clearer target for dismissal motions. Here, that process helped define which copyright theories would face early judicial testing.
One of the most closely watched rulings in the MDL came on October 27, 2025, when Judge Stein denied OpenAI’s motion to dismiss the consolidated class plaintiffs’ output-based direct copyright infringement claim. The court did not find liability, and it did not resolve the broader legality of AI training on copyrighted books. But it did allow that specific theory to move past the pleading stage.
The ruling focuses on whether the AI system’s generated responses can themselves support a direct infringement theory tied to protected works. Allowing that theory to proceed does not establish that the plaintiffs will ultimately prove copying, substantial similarity, causation, or recoverable harm. It does mean, however, that the court found the allegations sufficient at the pleading stage to justify further litigation on that issue.
Public filings in late 2025 and early 2026 show that the litigation has continued beyond the motion-to-dismiss stage and into active discovery disputes. Among the public issues addressed in the MDL were disputes concerning deleted datasets, privilege assertions, and the scope of discovery into how certain book-related datasets were handled. Those disputes underscore that this litigation will likely turn not only on legal doctrine, but also on technical records, internal communications, and evidentiary detail.
As of late March 2026, there has been no public final judgment, settlement, or dismissal ending the Authors Guild matter or the broader MDL. The case remains active. At this point, the dispute stands as an ongoing federal test of how copyright law may apply to books, AI model development, generated outputs, and commercial deployment in the generative AI setting.
These experts may assess alleged market harm, lost licensing opportunities, and the economic effect of unauthorized copying on authors and rights holders. In a case like this, their work may become especially important if plaintiffs seek to show that AI systems compete with or diminish established book-licensing markets.
These experts may explain how large language models are trained, how text data is processed, and what technical steps separate training inputs from generated outputs. Their analysis may help the court understand whether plaintiffs’ theories about copying and output generation are technically plausible.
These experts may examine how datasets were assembled, whether copyrighted works can be traced into development pipelines, and what internal records reveal about data handling. In litigation involving books and disputed datasets, that kind of analysis may become central during discovery.
These experts may address how book markets function, how licensing arrangements are typically structured, and how authors derive economic value from their works. Their opinions may help frame whether the alleged conduct caused cognizable harm in a real-world publishing context.
These experts may compare model outputs with copyrighted texts and assess whether any similarities are legally meaningful. In an output-based copyright theory, that analysis may become particularly important because plaintiffs must do more than raise general concerns about AI systems. They must connect specific outputs to protected expression in a way that supports an infringement claim.
The survival of the output-based direct infringement claim shows that courts may closely examine how specifically plaintiffs tie protected works to alleged outputs. In AI copyright cases, generalized allegations may not be enough. The pleadings must present a concrete theory that gives the court a workable legal framework.
This litigation reaches beyond the alleged ingestion of books into model development. It also addresses commercialization, deployment, and allegedly infringing outputs. That broader framing may influence both liability theories and the range of discovery sought from defendants.
Once related cases are centralized, individual complaints matter less than consolidated pleadings, coordinated motion practice, and shared discovery battles. That can increase efficiency, but it can also raise the stakes of each major ruling because one decision may influence a wider group of claims.
The public record already shows that discovery disputes are not peripheral. Questions about datasets, internal communications, deletion decisions, and privilege assertions may become highly significant. For counsel on either side, technical records may prove just as important as abstract copyright arguments.
Because the plaintiffs’ claims involve copyrighted books and commercial AI systems, questions about substitution, licensing, and economic effect are likely to stay at the center of the case. That is especially true if the litigation moves deeper into class issues, damages, or summary judgment.
The Authors Guild plaintiffs allege that copyrighted books were copied and used without permission in connection with commercial generative AI systems. OpenAI and Microsoft deny liability, and the court has not yet resolved the core copyright questions raised by the case. What is already clear, however, is that the dispute has developed into a substantial federal proceeding involving consolidated pleadings, consequential motion practice, and active discovery over how book-related materials were handled.
The case includes real rulings on pleading sufficiency, real disputes over technical evidence, and real pressure on courts to decide how traditional copyright principles apply to generative AI systems. That is why the case continues to matter not only to authors and technology companies, but also to litigators watching how federal courts respond when copyright law meets large-scale AI development.
When was Authors Guild et al. v. OpenAI Inc. filed?
The original complaint was filed on September 19, 2023, in the Southern District of New York.
Why was Microsoft added to the case?
Microsoft was added in the amended complaint filed on December 4, 2023. The plaintiffs alleged that Microsoft was involved in the deployment and commercialization of the AI systems identified in the lawsuit.
What changed when the case became part of the OpenAI copyright MDL?
Once the case was transferred into the MDL in April 2025, it became part of a coordinated federal structure handling overlapping copyright actions against OpenAI and, in some instances, Microsoft. That changed how pleadings, discovery disputes, and major pretrial issues were managed.
Why is the October 27, 2025 ruling important?
The ruling is important because the court allowed the plaintiffs’ output-based direct copyright infringement theory to move past the pleading stage. The decision did not determine liability, but it confirmed that this part of the case was sufficiently pleaded to continue in federal court.
What issues are likely to matter as the litigation continues?
Likely issues include the handling of datasets, the connection between protected books and model outputs, alleged market harm to authors, and the evidence needed to support or defeat copyright claims as the case moves through discovery and later motion practice.
These experts may assess alleged market harm, lost licensing opportunities, and the economic effect of unauthorized copying on authors and rights holders. In a case like this, their work may become especially important if plaintiffs seek to show that AI systems compete with or diminish established book-licensing markets.
These experts may explain how large language models are trained, how text data is processed, and what technical steps separate training inputs from generated outputs. Their analysis may help the court understand whether plaintiffs’ theories about copying and output generation are technically plausible.
These experts may examine how datasets were assembled, whether copyrighted works can be traced into development pipelines, and what internal records reveal about data handling. In litigation involving books and disputed datasets, that kind of analysis may become central during discovery.
These experts may address how book markets function, how licensing arrangements are typically structured, and how authors derive economic value from their works. Their opinions may help frame whether the alleged conduct caused cognizable harm in a real-world publishing context.
These experts may compare model outputs with copyrighted texts and assess whether any similarities are legally meaningful. In an output-based copyright theory, that analysis may become particularly important because plaintiffs must do more than raise general concerns about AI systems. They must connect specific outputs to protected expression in a way that supports an infringement claim.
The survival of the output-based direct infringement claim shows that courts may closely examine how specifically plaintiffs tie protected works to alleged outputs. In AI copyright cases, generalized allegations may not be enough. The pleadings must present a concrete theory that gives the court a workable legal framework.
This litigation reaches beyond the alleged ingestion of books into model development. It also addresses commercialization, deployment, and allegedly infringing outputs. That broader framing may influence both liability theories and the range of discovery sought from defendants.
Once related cases are centralized, individual complaints matter less than consolidated pleadings, coordinated motion practice, and shared discovery battles. That can increase efficiency, but it can also raise the stakes of each major ruling because one decision may influence a wider group of claims.
The public record already shows that discovery disputes are not peripheral. Questions about datasets, internal communications, deletion decisions, and privilege assertions may become highly significant. For counsel on either side, technical records may prove just as important as abstract copyright arguments.
Because the plaintiffs’ claims involve copyrighted books and commercial AI systems, questions about substitution, licensing, and economic effect are likely to stay at the center of the case. That is especially true if the litigation moves deeper into class issues, damages, or summary judgment.
The Authors Guild plaintiffs allege that copyrighted books were copied and used without permission in connection with commercial generative AI systems. OpenAI and Microsoft deny liability, and the court has not yet resolved the core copyright questions raised by the case. What is already clear, however, is that the dispute has developed into a substantial federal proceeding involving consolidated pleadings, consequential motion practice, and active discovery over how book-related materials were handled.
The case includes real rulings on pleading sufficiency, real disputes over technical evidence, and real pressure on courts to decide how traditional copyright principles apply to generative AI systems. That is why the case continues to matter not only to authors and technology companies, but also to litigators watching how federal courts respond when copyright law meets large-scale AI development.
When was Authors Guild et al. v. OpenAI Inc. filed?
The original complaint was filed on September 19, 2023, in the Southern District of New York.
Why was Microsoft added to the case?
Microsoft was added in the amended complaint filed on December 4, 2023. The plaintiffs alleged that Microsoft was involved in the deployment and commercialization of the AI systems identified in the lawsuit.
What changed when the case became part of the OpenAI copyright MDL?
Once the case was transferred into the MDL in April 2025, it became part of a coordinated federal structure handling overlapping copyright actions against OpenAI and, in some instances, Microsoft. That changed how pleadings, discovery disputes, and major pretrial issues were managed.
Why is the October 27, 2025 ruling important?
The ruling is important because the court allowed the plaintiffs’ output-based direct copyright infringement theory to move past the pleading stage. The decision did not determine liability, but it confirmed that this part of the case was sufficiently pleaded to continue in federal court.
What issues are likely to matter as the litigation continues?
Likely issues include the handling of datasets, the connection between protected books and model outputs, alleged market harm to authors, and the evidence needed to support or defeat copyright claims as the case moves through discovery and later motion practice.
Copyright Infringement
Publishing
Licensing
Plagiarism
Computer Linguistics
Data Mining Algorithm
Technology
Patent, +4
Technology
Artificial Intelligence, +4
Technology