Meta Sued by Hundreds of Authors Over AI Training, 292 Plaintiffs, Llama and Muse Claims, $150,000 Damages and What Happens Next

On October 19, 2026, hundreds of book authors filed new copyright lawsuits against Meta Platforms in the Northern District of California. The authors say Meta copied their books, including from pirate “shadow libraries,” and used them to train its Llama and newer Muse AI models. One theory goes further than earlier cases: it challenges Meta’s scanning of physical books it owned.

These are not class actions, and nothing is open to claim. Each suit names a group of individual authors as plaintiffs. There is no settlement and no claim form, and the allegations have not been proven. Meta had not responded in court to the new complaints at the time of writing.

Quick Facts: Authors v. Meta AI Training Lawsuits

Detail	Information
Defendant	Meta Platforms, Inc.
Filed	October 19, 2026
Court	U.S. District Court, Northern District of California
Number of authors	About 300. One litigation tracker counts 292 across five suits. Bloomberg Law reported four cases.
Cases named in reporting	Winspear, Sheard, Older, Reno and (per the tracker) Levine, each v. Meta
Plaintiffs’ firms	Freedman Normand Friedland and Stris & Maher
AI models at issue	Llama and Muse
Lead plaintiff in Winspear	Jacqueline Winspear, author of the Maisie Dobbs series
Type of case	Multi-plaintiff individual suits, not a class action
More suits expected?	The attorneys said they plan claims for nearly 1,000 more authors in the coming months
Meta liable?	No finding. The complaints are allegations
Settlement or claim form	None

The Cases

Reporting on the filings differs slightly. A litigation tracker counted five near-identical suits covering 292 authors. Bloomberg Law reported four cases and about 300 authors. The most likely explanation is a difference in which cases were counted, but check the dockets for the final list.

Case	Authors (tracker count)	Case number (Bloomberg Law)
Winspear v. Meta Platforms	60	3:26-cv-12100
Sheard v. Meta Platforms	58	3:26-cv-12126
Older v. Meta Platforms	57	3:26-cv-12115
Levine v. Meta Platforms	61	Not in the Bloomberg list
Reno v. Meta Platforms	56	3:26-cv-12131
Total	292

The five complaints reportedly make nearly identical allegations, so consolidation before Judge Vince Chhabria, who handles other book cases against Meta, is expected. That is a prediction, not a ruling. Courts can coordinate related cases under Federal Rule of Civil Procedure 42, but consolidation is not automatic. Until an order is entered, treat each as a separate action.

What the Authors Allege

According to the reported complaints:

• Shadow-library downloads. Meta allegedly obtained electronic copies of the authors’ books, in the hundreds of thousands of titles overall, from unauthorized repositories and used them to train large language models without permission.
• Torrenting and distribution. Downloading through torrent systems can involve uploading to others, which implicates the exclusive right to distribute. This issue remained open in the earlier Kadrey v. Meta case.
• Llama and Muse. The suits reach beyond the older Llama models to Meta’s newer Muse models.
• Scanning physical books (“Project Vellum”). The newer theory is that even scanning physical books Meta lawfully owned was infringement. Meta disclosed Project Vellum in discovery in a related case, Chicken Soup for the Soul v. Meta, No. 3:26-cv-02333-VC (N.D. Cal.).

The authors do not necessarily have to prove that Meta’s AI reproduces a book word for word. A copyright claim can rest on unauthorized copying during training. Meta can still raise defenses, including fair use.

What Copyright Law Protects

Under 17 U.S.C. § 106, copyright owners have exclusive rights to reproduce, adapt and distribute their works, subject to limits such as fair use. Copyright protects an author’s original expression, such as the wording of a novel, not ideas, facts or genres.

Anyone who violates those rights is an infringer under 17 U.S.C. § 501. Before suing, a U.S. work generally must be registered with the Copyright Office (17 U.S.C. § 411(a)), a rule the Supreme Court confirmed in Fourth Estate v. Wall-Street.com (2019). The usual time limit to sue is three years (17 U.S.C. § 507(b)).

Could Meta Owe $150,000 Per Book?

Potentially, but that is a ceiling, not a prediction.

• The ranges. Under 17 U.S.C. § 504(c), statutory damages are generally $750 to $30,000 per infringed work. If the court finds infringement was willful, it may raise that to $150,000 per work. For innocent infringement it can be as low as $200.
• Per work, not per copy. The award is calculated per work, not per download, training run or model version.
• Registration timing. Under 17 U.S.C. § 412, statutory damages and attorney’s fees generally are unavailable for infringement that began before registration, unless the work was registered within three months of first publication. An author can own a valid copyright and still lose access to statutory damages.
• Willfulness must be proven. An allegation of intentional copying does not establish willfulness.
• No total can be calculated yet. Multiplying 300 authors by $150,000 per book is not a realistic estimate.

Other remedies can include actual damages and profits, an injunction and, where allowed, attorney’s fees and costs (17 U.S.C. §§ 502, 504 and 505).

Why Authors Are Suing Individually Instead of as a Class

Class payouts in book cases have been estimated at around $3,000 per title. Statutory damages can range higher, which is one reason some authors prefer individual suits. The tradeoffs are real:

• Each plaintiff must prove registration, ownership and infringement for their own works.
• Individual suits are more expensive, and a loss on fair use leaves the plaintiffs with nothing.
• Different authors have different registration dates, contracts and evidence, which can also make class treatment harder.

Choosing this strategy does not show the claims are stronger. A class action under Federal Rule of Civil Procedure 23 requires common questions and adequate representation, while individual claims keep author-specific questions in play.

Related article: Hospital Español Auxilio Mutuo $500,000 Data Breach Settlement, Claim $25 or Up to $2,500 by December 21, 2026 (Morales Rosario v. Hospital Español Auxilio Mutuo de Puerto Rico, Inc., No. 3:25-cv-01244-RAM)

Meta’s Main Defense: Fair Use

Fair use, 17 U.S.C. § 107, will be central. Courts weigh four factors, and no single rule makes all AI training lawful or unlawful.

1. Purpose and character. Meta can argue that training is transformative because the model learns language patterns rather than distributing copies of a book. The authors can respond that the books were copied to build a commercial product without a license.
2. Nature of the work. Fiction and other creative works receive stronger protection than purely factual material.
3. Amount copied. Training datasets can contain complete books. Copying a whole work does not automatically defeat fair use but weighs on the analysis.
4. Market effect. This is the major battleground. Authors argue AI can substitute for books or undercut licensing markets. AI developers argue training serves a different purpose. Courts want evidence.

What Earlier Rulings Say

Bartz v. Anthropic (June 2025). Judge William Alsup held that training on lawfully acquired books was fair use, and that converting purchased print books into digital copies was also fair use. He held that building a library of pirated books was not. That split is why the “Project Vellum” scanning claim faces an uphill argument, and why the shadow-library allegations are the authors’ stronger ground. The case settled for $1.5 billion in a class action, and reports say the court approved the deal in July 2026. A settlement in one case does not establish liability for another company.

Kadrey v. Meta (June 2025). Judge Chhabria granted Meta summary judgment on the training claim for 13 authors, including Richard Kadrey, Sarah Silverman and Ta-Nehisi Coates, because they had not shown market harm. He stressed the ruling did not mean Meta’s use of copyrighted books is lawful in general, and he discussed “market dilution” as a theory that might succeed with better evidence. The torrenting and distribution question was not decided. Reports say he later denied a request to fast-track an appeal.

Thomson Reuters v. Ross (D. Del. 2025). A court rejected fair use for a non-generative AI tool trained on copied headnotes, showing that courts are not uniform.

Does Kadrey defeat the new claims? No. It is binding only in that case and rested on that record. But it gives Meta a strong argument, and the new plaintiffs will likely try to build the market-harm evidence the Kadrey plaintiffs lacked.

Other Meta and AI Copyright Cases

• Publisher and author suit. In May 2026, Elsevier, Cengage, Hachette, Macmillan and McGraw Hill, with author Scott Turow, sued Meta and Mark Zuckerberg, alleging pirated books and journal articles were used to train Llama. For our coverage, see the Meta Llama publisher copyright class action.
• Textbook authors. A proposed class action filed July 2, 2026 alleges Meta pirated textbooks, removed copyright-management information, and shared files while torrenting. Removing that information can be a separate violation under 17 U.S.C. § 1202.
• Anthropic. For the approved class settlement, see Anthropic’s $1.5 billion copyright settlement.
• Nvidia. In Nazemian v. NVIDIA, a federal judge reportedly allowed authors’ direct and contributory infringement claims over Books3 data to proceed in May 2026. See the Nvidia NeMo copyright lawsuit.
• Scale. One unofficial tracker counts 17 AI-related copyright suits against Meta and roughly 150 against AI companies in the U.S., though some overlap or are consolidated.

Each case has its own parties, evidence and rulings. None automatically decides the others.

What Happens Next

Likely steps include service of the complaints, Meta’s response (an answer or motions to dismiss), possible coordination of the cases, discovery into datasets, copying, training and market effects, disputes over which works qualify, and then settlement, motions for judgment or trial. No schedule is confirmed. Because these are new filings, a resolution is likely years away.

What This Means for Authors

• Nothing is open to claim. There is no settlement or claim form. Do not pay anyone who promises a guaranteed Meta payout.
• Gather your records. Useful items include copyright registration certificates, publication dates, publishing contracts and rights reversion or licensing agreements, and any evidence your book was in a dataset. A book’s appearance in a dataset does not by itself prove every element of a claim.
• Check registration. Registration timing can decide whether statutory damages and fees are available. Registering now does not retroactively fix earlier infringement, but it can matter for later claims.
• Know who owns your rights. If you assigned rights to a publisher or licensed AI uses, the contract may determine who can sue. Self-published authors who own their rights may have the most direct claim.
• Watch for class cases. If you could be a member of a proposed class, such as the textbook authors case, deadlines to opt out or object can affect your right to sue separately. Follow those dockets.
• Talk to a copyright attorney. Whether to join a suit, wait for a class or do neither depends on your works and contracts.

What This Means for Users of Meta Platforms

These are copyright claims by authors and publishers. They do not, by themselves, give ordinary Facebook or Instagram users a claim, and they do not create a consumer compensation program. Privacy, data and content-moderation disputes follow different laws.

Frequently Asked Questions

How many authors are suing Meta?

About 300. A litigation tracker counted 292 across five suits, and Bloomberg Law reported about 300 across four cases. The attorneys say nearly 1,000 more authors may follow.

Are these class actions?

No. They are suits by named groups of individual authors. The publisher suit and textbook authors suit are proposed class actions.

What does the complaint say Meta did?

That it copied books from unauthorized online libraries and used them to train Llama and Muse without permission, and that scanning physical books was also infringement. Meta has not been found liable.

What is Project Vellum?

It is the name for Meta’s program of scanning physical books, disclosed in discovery in a related case. The authors say that scanning was infringement. Another federal judge has held that scanning purchased print books was fair use.

Which judge will hear the cases?

They were filed in the Northern District of California and are expected to go to Judge Vince Chhabria, who handles the other book cases against Meta. Assignment is not confirmed.

How much money could the authors get?

Statutory damages are generally $750 to $30,000 per registered work, up to $150,000 if infringement is willful. Those are ceilings, not predictions, and no award has been made.

Did Meta already win an AI copyright case?

Meta won summary judgment in Kadrey for 13 authors in June 2025. The judge limited the decision to that record and did not hold Meta’s use lawful in general.

Does the Anthropic $1.5 billion settlement mean Meta must pay?

No. It resolved claims against Anthropic and does not establish liability or damages against Meta.

Can I join the lawsuits?

There is no open claims program. The attorneys have said they plan more filings, but joining depends on your works, registrations and contracts. Speak to a copyright attorney.

Can a self-published author sue?

Possibly, if the author owns the relevant rights and the legal requirements, including registration, are met.

Is there a claim form?

No court-approved claim form or compensation program has been identified as of October 20, 2026.

Where can I follow updates?

Check each case’s docket on PACER and the Northern District of California’s website. The complaints and court orders are the authoritative sources.

Sources

1. Bloomberg Law, “Meta Sued by Hundreds of Authors in AI Training Copyright Cases” (October 20, 2026).
2. Litigation tracker coverage of Winspear, Sheard, Older, Levine and Reno v. Meta (October 19, 2026).
3. Chicken Soup for the Soul, LLC v. Meta Platforms, Inc., No. 3:26-cv-02333-VC (N.D. Cal.).
4. Kadrey v. Meta Platforms, Inc. (N.D. Cal. June 2025) and Bartz v. Anthropic PBC (N.D. Cal. June 2025), as reported by the Associated Press and WIPO Lex.
5. Reuters coverage of the approval of the Anthropic settlement (July 2026).
6. Book Riot and Publishers Weekly coverage of the May 2026 publishers’ suit against Meta.
7. Textbook Authors Association, “Textbook Authors File Lawsuit Against Meta” (July 2026).
8. 17 U.S.C. §§ 106, 107, 411, 412, 501, 502, 504, 505, 507 and 1202.
9. Fourth Estate Public Benefit Corp. v. Wall-Street.com, LLC, 586 U.S. 296 (2019); Thomson Reuters v. Ross Intelligence (D. Del. 2025).
10. U.S. Copyright Office, “Copyright and Artificial Intelligence” (copyright.gov/ai).
11. U.S. District Court, Northern District of California (cand.uscourts.gov).

Written by Israr Ahmad, legal content researcher at AllAboutLawyer.com. Last updated October 20, 2026. This article is general information, not legal advice, and AllAboutLawyer.com is not a law firm. The complaints contain allegations that have not been proven. Damages and defenses depend on the works, ownership, registration, evidence and the court’s rulings. No settlement or claim form has been identified for the new cases.