AI Training Data Copyright: Fair Use, Licensing, and Infringement Risk

The legal question of whether training AI models on copyrighted works constitutes infringement has moved from academic debate to courtroom reality. With billions of dollars in damages at stake and several cases now decided at summary judgment, AI companies face immediate decisions about data sourcing, licensing strategy, and risk mitigation. The early rulings do not point in one direction: courts have treated model training itself as strongly transformative while drawing a hard line at the use of pirated source copies.

AI Training Data Copyright: The Short Answer

The central AI training data copyright question is no longer whether copying occurs—building a training dataset implicates the reproduction right—but whether fair use excuses it. The 2025 rulings split: courts found training a general-purpose model strongly transformative, yet Bartz v. Anthropic held that downloading and retaining pirated source copies is infringing,¹ the holding behind Anthropic’s $1.5 billion settlement.² Outcomes turn on the specific use and the specific record, and a developer who trains on lawfully obtained materials stands on far firmer ground than one who builds its corpus from pirated libraries.

This guide provides a practical framework for understanding copyright risks in AI training, evaluating fair use defenses, implementing compliant data sourcing practices, and navigating the evolving regulatory landscape.

The Legal Framework: Copyright Law and AI Training

Does Training on Copyrighted Data Constitute Infringement?

In the Office’s view, expressed in its May 2025 pre-publication report, building a training dataset using copyrighted works “clearly implicate[s] the right of reproduction”—making the act of copying presumptively infringing unless a defense like fair use applies.³ That framing is influential but not settled law: the document is a pre-publication version rather than a final report. The practical point holds regardless: copying copyrighted works into a training dataset requires either authorization or a valid legal defense.

The copyright infringement analysis follows established principles:

What constitutes copying: Copying copyrighted works into a training dataset—during collection, preprocessing, and storage—exercises the copyright owner’s exclusive right “to reproduce the copyrighted work in copies or phonorecords.”⁴

Rights holder standing: Copyright owners can sue for infringement even if their works represent a tiny fraction of a massive training dataset. Class action lawsuits aggregate claims from thousands of copyright holders, multiplying potential exposure.

Chain of liability: Companies face liability not only for works they directly copied, but also for works obtained from third-party datasets if those datasets were assembled through unauthorized copying (such as the LAION-5B dataset built from scraped images).

The Four-Factor Fair Use Test

Fair use provides the primary defense for unauthorized use of copyrighted training data. Courts evaluate the four statutory factors set out in 17 U.S.C. § 107:⁵

Factor 1: Purpose and Character of Use

This factor examines whether the use is “transformative”—whether it adds new meaning, expression, or purpose rather than merely superseding the original work.

In the 2025 summary-judgment rulings, courts have found that training a general-purpose model on text can be strongly transformative—one judge described it as “exceedingly transformative.”⁶ The Copyright Office’s pre-publication analysis draws a related distinction:

• Most transformative: Training a model to deploy it for research, “or in a closed system that constrains it to a non-substitutive task”⁷
• Less transformative: Training a model to generate outputs “substantially similar to copyrighted works in the dataset,” or training it “on specific types of works in order to produce content that shares the purpose of appealing to a particular audience”⁷

The Copyright Office cautioned against treating AI training as inherently transformative simply because it involves computational analysis rather than human reading, noting that AI systems create perfect copies with the ability to analyze works nearly instantaneously.⁸ Early case law has been more receptive to the transformative-use argument for the training step itself, while courts have separately scrutinized how the underlying copies were obtained.

Commercial purpose weighs against fair use but is not dispositive if other factors favor the defendant.

Factor 2: Nature of the Copyrighted Work

This factor considers the type of work copied:

• Highly expressive creative works (novels, artwork, music, photography): Courts afford these stronger protection, disfavoring fair use
• Factual or functional works (databases, technical documentation, news articles): These receive thinner protection, favoring fair use
• Published vs. unpublished works: Courts treat unpublished status as weighing against fair use, because the author retains the right of first publication, though the statute adds that “[t]he fact that a work is unpublished shall not itself bar a finding of fair use if such finding is made upon consideration of all the above factors.”⁹

Most AI training involves published, creative works—a combination that ordinarily weighs against fair use, though it has not proven decisive in the recent text-training rulings.

Factor 3: Amount and Substantiality of Use

AI training typically involves copying entire works rather than excerpts or portions. While wholesale copying ordinarily weighs against fair use, the Copyright Office acknowledged that copying complete works “may be necessary” for certain types of training.¹⁰ Its qualifier matters for Factor 3: the Office reasoned that while “for large, general-purpose models, there is no need to copy any amount of any specific work,” research supports the assertion that “internet-scale pre-training data, including large amounts of entire works, may be necessary to achieve the performance of current-generation models.”¹⁰ A necessity argument therefore runs to the corpus in the aggregate, not to any particular plaintiff’s work.

Courts will evaluate whether copying entire works was reasonably necessary to achieve the transformative purpose, or whether training on smaller portions or samples would have sufficed.

Factor 4: Effect on the Market

The Copyright Office broadly interpreted this factor to encompass:

• Direct substitution: Whether AI-generated outputs replace sales of original works
• Market dilution: Whether proliferation of AI-generated content in similar styles undermines market value even without direct substitution
• Lost licensing opportunities: Whether the unauthorized use deprives copyright holders of potential licensing revenue¹¹

The Copyright Office reads the fourth factor to disfavor unlicensed use where licensing markets are available to meet training needs, while noting that where barriers to licensing prove insurmountable there may be no functioning market to harm.¹¹ As licensing marketplaces proliferate and major publishers negotiate training data deals, courts may view unauthorized use as market displacement rather than fair use.

This factor has emerged as the decisive battleground. In Kadrey v. Meta, the court treated market dilution as the dispositive question and granted the defendant summary judgment only because the plaintiffs failed to develop a market-harm record—while signaling that a better-supported dilution theory could prevail in a future case.¹² The lesson for both rights holders and developers is that Factor 4 outcomes are record-dependent, not foreordained.

Current Litigation Landscape: What Courts Are Deciding

The 2025 summary-judgment rulings do not line up neatly for either side. One court rejected fair use for a non-generative legal-research tool; two others found model training transformative while splitting on how the source copies were acquired and whether market harm was proven. The throughline is that outcomes turn on the specific use and the specific record, not on a categorical rule that AI training is or is not infringement. Side by side, the three summary-judgment rulings split both on the training step and on what ultimately decided each case.

The June 2025 Fair Use Rulings: A Split Emerges

Two decisions issued days apart in the Northern District of California reframed the field, both finding the training step transformative while diverging on the issues that decided each case.

Bartz v. Anthropic PBC, 787 F. Supp. 3d 1007 (N.D. Cal. June 23, 2025) (Alsup, J.): The court issued a split summary-judgment ruling. It held that using books to train large language models is “exceedingly” and “spectacularly” transformative and therefore fair use, and that digitizing lawfully purchased print books (a format change, with the originals discarded) was also fair use. But it held that downloading and retaining pirated copies—more than seven million books pulled from pirate libraries to build a permanent central library—was not fair use and constituted infringement, setting the piracy for trial on liability and damages.¹³

Kadrey v. Meta Platforms, Inc., 788 F. Supp. 3d 1026 (N.D. Cal. June 25, 2025) (Chhabria, J.): The court granted Meta summary judgment on the training claim, finding that training its models on books (including pirated copies) was highly transformative. The ruling turned on Factor 4: the plaintiffs failed to prove market dilution, which the court treated as dispositive on this record. The court was careful to add that this outcome reflected a weak record, not a safe harbor—signaling that a well-developed market-dilution theory could succeed in a future case. On the question the opinion framed at the outset—whether feeding copyright-protected material into a model without permission is illegal—the court wrote that “[a]lthough the devil is in the details, in most cases the answer will likely be yes.”¹⁴

Implications: Read together, these rulings establish that the training step has meaningful fair use support, but that the provenance of the source copies and the strength of the market-harm record can be decisive. A developer who trains on lawfully obtained materials stands on far firmer ground than one who builds its corpus from pirated libraries.

Thomson Reuters v. Ross Intelligence: Fair Use Rejected for a Competing Research Tool

In February 2025, a Delaware federal court issued an early decision rejecting fair use for AI training data. Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc., 765 F. Supp. 3d 382 (D. Del. Feb. 11, 2025) (Bibas, J., sitting by designation). Thomson Reuters sued Ross Intelligence for training a legal research tool on Thomson Reuters’s proprietary Westlaw headnotes without authorization. The ruling revised the court’s own 2023 order, which had sent the fair use question to a jury.¹⁵

Key holdings:

• Not transformative: The court found Ross’s use “not transformative” because the tool served the same purpose as Westlaw’s original content—legal research. Despite the technological sophistication, the functional purpose remained identical.¹⁶
• Market harm: The court called this factor “undoubtedly the single most important element of fair use,” finding that Ross’s product directly competed with Westlaw and could harm both the existing market for legal-research platforms and the potential derivative market for data to train legal AIs.¹⁷
• Direct infringement: The court granted partial summary judgment on direct copyright infringement.¹⁸

Posture and limits: The ruling is now on interlocutory appeal to the Third Circuit, so it is best read as persuasive but not settled.¹⁹ Its reach is also narrower than it first appears: Ross’s tool was a non-generative legal-research product trained on Westlaw headnotes, and the court emphasized that Ross’s output competed directly with the source. Companies building generative models on broad, diverse corpora should be cautious about over-reading Ross, whose facts differ materially from the foundation-model context.

Andersen v. Stability AI: Artists’ Case Proceeds to Trial

In August 2024, Judge William Orrick allowed the visual artists’ copyright claims to proceed against Stability AI, Midjourney, DeviantArt, and Runway AI, while dismissing their DMCA claims with prejudice. Andersen v. Stability AI Ltd., 744 F. Supp. 3d 956 (N.D. Cal. Aug. 12, 2024).²⁰ The case is proceeding, with a jury trial scheduled for September 20, 2027.²¹

Key findings:

• Plausible infringement claims: The court found that the artists plausibly alleged Stable Diffusion was built “to a significant extent on copyrighted works” and was “created to facilitate that infringement by design.”²²
• Storage theory: At the pleading stage the court held that plaintiffs’ works being contained in Stable Diffusion “as algorithmic or mathematical representations”—a different medium than the originals—“is not an impediment to the claim at this juncture,” expressly reserving the model and distribution theories for summary judgment.²³
• Discovery phase: The case proceeding to discovery could expose internal communications about training data sourcing and corporate decision-making regarding copyright compliance.

Implications: At the pleading stage, this decision allowed the theory to proceed that training AI on scraped artwork without authorization may constitute infringement, even when the model doesn’t store pixel-perfect copies.

OpenAI Litigation: Consolidated in an MDL

OpenAI faces numerous copyright lawsuits from authors, news publishers, and other content creators:

Authors’ lawsuits: Sarah Silverman, Paul Tremblay, Ta-Nehisi Coates, Michael Chabon, and other authors allege OpenAI trained ChatGPT on their copyrighted books without permission.

The New York Times case: The Times seeks “billions of dollars” in damages for unauthorized use of articles to train GPT models.²⁴ Earlier in the litigation, the court denied OpenAI’s motion to compel evidence about the Times’s own use of generative AI.²⁵

Current posture: The U.S. cases have been consolidated in a multidistrict litigation—In re OpenAI, Inc., Copyright Infringement Litigation, MDL No. 3143, centralized by In re OpenAI, Inc., 776 F. Supp. 3d 1352 (J.P.M.L. Apr. 3, 2025) before Judge Sidney H. Stein in the Southern District of New York.²⁶ Motions to dismiss have been decided case by case rather than MDL-wide: in the Ziff Davis action, for example, the court dismissed two claims, dismissed a third in part, and denied the motion as to three others. Davis v. OpenAI (In re OpenAI, Inc. Copyright Infringement Litig.), 2025 U.S. Dist. LEXIS 259061 (S.D.N.Y. Dec. 15, 2025).²⁷ On the discovery side the court directed OpenAI to produce twenty million de-identified consumer ChatGPT output logs, In re OpenAI, Inc., 2025 U.S. Dist. LEXIS 222099 (S.D.N.Y. Nov. 7, 2025) (Wang, M.J.),²⁸ and then denied OpenAI’s motion to reconsider that order and directed production within seven days of completing de-identification, In re OpenAI, Inc., 2025 U.S. Dist. LEXIS 252574 (S.D.N.Y. Dec. 2, 2025)²⁹—a development with significant implications for proving, or rebutting, infringement and output similarity.

International suits: Parallel suits have followed abroad, including in Canada and India.

GitHub Copilot: Code Training Under Scrutiny

Developers sued GitHub, Microsoft, and OpenAI in November 2022, alleging Copilot was trained on billions of lines of open-source code without complying with licensing terms.

Case status: The court (Judge Jon Tigar, N.D. Cal., No. 4:22-cv-06823) dismissed most claims with leave to amend but denied the motions to dismiss the claim for breach of the open-source licenses and the DMCA §§ 1202(b)(1) and (b)(3) claims. Doe v. GitHub, Inc., 672 F. Supp. 3d 837 (N.D. Cal. May 11, 2023):³⁰

• Breach of the open-source licenses (a contract claim), which the court held was adequately pleaded
• Claims under DMCA §§ 1202(b)(1) and 1202(b)(3), as to which the court denied the motions to dismiss
• Claims for violation of DMCA §§ 1202(a) and 1202(b)(2), tortious interference, fraud, false designation of origin, unjust enrichment, unfair competition, breach of the GitHub Privacy Policy and Terms of Service, the CCPA, and negligence, all dismissed with leave to amend
• Claims for civil conspiracy and declaratory relief, dismissed with prejudice

In a later order, the court dismissed the plaintiffs’ surviving DMCA Section 1202(b)³¹ claims with leave to amend,³² and on June 24, 2024 it dismissed the repleaded claims. The court certified that dismissal for interlocutory appeal and stayed its proceedings, and the Ninth Circuit granted permission to appeal.³³

On September 16, 2026, the Ninth Circuit affirmed in a published opinion. The panel held that the plaintiffs had standing but did not state a Section 1202(b) claim: on their own allegations, Copilot “is best understood as learning from existing works and then creating new works based on that learning process, not as making copies of existing works,” so it does not “remove or alter” copyright management information from a copy of an existing work. The court called “identicality” “something of a misnomer,” “best understood as a gloss on the statutory terms ‘remove,’ ‘alter,’ and ‘copies’ rather than an independent (and atextual) element of a section 1202(b) claim,” declined as forfeited to consider a training-stage “input” theory, and expressed “no view on whether that similarity would allow plaintiffs to assert a claim for copyright infringement.” The breach-of-license claims remain pending in the district court.³⁴

Key issue: Whether an AI tool that reproduces licensed code as output without attribution, copyright notice, or license terms breaches the open-source licenses that supplied the code.

Implications: Even permissive open-source licenses impose conditions—attribution, copyright notice, and license terms—that an AI tool’s output may violate. Companies training on open-source code must evaluate license compliance, not just copyright law.

Anthropic’s $1.5 Billion Settlement: Resolving the Piracy Claim

In September 2025, Anthropic agreed to pay $1.5 billion to settle authors’ claims—approximately $3,000 per work across the 482,460 works on the final works list.³⁵ The settlement resolves the piracy liability that Judge Alsup found in Bartz; it does not resolve the fair use question on the training itself, which Anthropic won.

The court held two hearings on the proposed settlement, on September 8 and September 25, 2025; the second ended in preliminary approval, memorialized in an opinion of October 17, 2025 that set the fairness hearing for April 23, 2026.³⁶ The final-approval motions were heard on May 14, 2026, and the court granted final approval and entered judgment on July 20, 2026.³⁷ The settlement caps any fee request at 25 percent before costs and expenses, and class counsel’s fee is paid in installments proportionate to Anthropic’s payments into the fund, none of it before final approval.³⁸ The court awarded $101,561,111 in fees, nearly 6.8 percent of the fund, rejecting the 12.5 percent request, and found that 91.3 percent of the works list had been claimed as of April 16, 2026.³⁹

Significance:

• Damages signal: The settlement amount—four times the statutory minimum of $750 per work—reflects market expectations for copyright liability even in negotiated resolutions, particularly where pirated source material is involved.⁴⁰
• Potential exposure: Before settlement, Anthropic itself told the court that a class trial on the pirated copies would expose it “to potentially hundreds of billions of dollars of liability.”⁴¹
• Scope of resolution: Because the settlement addresses the piracy claim rather than the fair use of training, it confirms that how a company acquires its source copies can drive liability even where the training itself is defensible.

Risk Assessment: Evaluating Copyright Exposure by Data Source

Not all training data carries equal copyright risk. Companies should assess exposure based on source, authorization status, and intended use.

High-Risk Data Sources

Web-scraped content without permission:

• Risk level: Highest
• Why: Direct copying of copyrighted works without authorization or license
• Examples: Scraping news articles, blog posts, creative writing, artwork, photographs from websites
• Current status: Multiple lawsuits target this practice; courts viewing skeptically

Pirated or illegally obtained works:

• Risk level: Highest (willful infringement)
• Why: Copyright law provides enhanced damages (up to $150,000 per work) for willful infringement⁴²
• Examples: Books obtained from shadow libraries (Library Genesis, Sci-Hub), leaked datasets
• Current status: Bartz v. Anthropic held that downloading and retaining pirated copies is not fair use even where the training itself is transformative, and that holding underlies the $1.5 billion Anthropic settlement⁴³

Licensed works used beyond scope:

• Risk level: High
• Why: Breach of contract claims plus copyright infringement if license doesn’t permit AI training
• Examples: Subscribing to database for research but using it for commercial AI training
• Current status: Thomson Reuters v. Ross illustrates that copying routed through a third-party data vendor does not insulate the buyer: Thomson Reuters refused to license Westlaw to Ross because Ross was a competitor, Ross bought training data that LegalEase built from Westlaw headnotes, and the court granted summary judgment against Ross on fair use and partial summary judgment on direct infringement⁴⁴

Moderate-Risk Data Sources

Public datasets of uncertain provenance:

• Risk level: Moderate to High
• Why: Many popular datasets (CommonCrawl, LAION-5B) contain copyrighted works obtained through scraping
• Examples: Image-text pairs, web corpora, code repositories
• Current status: Downstream users may face liability for infringement in dataset assembly
• Mitigation: Investigate dataset creation methodology; prefer datasets with documented copyright clearance

Open-source code with restrictive licenses:

• Risk level: Moderate
• Why: Many open-source licenses require attribution, notices, or share-alike provisions that AI training may violate
• Examples: GPL, AGPL, Creative Commons ShareAlike (CC BY-SA)
• Current status: The GitHub Copilot litigation tests whether reproducing licensed code as output “without attribution, copyright notice, or license terms” breaches the open-source licenses; those claims remain pending after the Ninth Circuit’s September 16, 2026 affirmance of the DMCA dismissal⁴⁵
• Mitigation: Implement license compliance tracking; provide attribution mechanisms

User-generated content from platforms:

• Risk level: Moderate
• Why: Platform terms may authorize AI training, but users retain copyright in their content
• Examples: Reddit posts, Stack Overflow answers, social media content
• Current status: Platform terms protect the platform, not necessarily downstream AI developers
• Mitigation: Review platform terms; consider direct user consent mechanisms

Lower-Risk Data Sources

Licensed commercial datasets:

• Risk level: Low to Moderate
• Why: Explicit license authorization reduces copyright risk, though license compliance remains essential
• Examples: Shutterstock AI licensing, publisher agreements, specialized data vendors
• Cost: Substantial (millions to hundreds of millions for large-scale training)
• Mitigation: Negotiate broad license terms covering training, fine-tuning, and commercial deployment

Public domain works:

• Risk level: Low
• Why: No copyright protection eliminates infringement liability
• Examples: Works published 1930 and earlier (pre-1931), U.S. government works, Creative Commons Zero (CC0) works
• Limitations: Public domain datasets don’t represent contemporary culture; insufficient for most commercial applications
• Mitigation: Verify public domain status; beware of copyright restoration for foreign works

Permissively licensed content:

• Risk level: Low
• Why: Licenses explicitly permit broad use, including commercial applications
• Examples: Creative Commons Attribution (CC BY), MIT License, Apache License 2.0
• Limitations: Still requires license compliance (attribution, notices)
• Mitigation: Implement attribution systems; maintain license records

Company-owned or commissioned content:

• Risk level: Minimal
• Why: Company owns copyright or has explicit authorization from creators
• Examples: Internal documents, commissioned works with work-for-hire agreements
• Limitations: Volume typically insufficient for foundation model training
• Mitigation: Document ownership or assignment agreements

Special Consideration: Synthetic Data

Risk profile: Variable, requires careful analysis

Synthetic data—artificially generated by AI models rather than collected from real-world sources—has been promoted as a copyright-safe alternative. However, synthetic data solves less of the copyright problem than it appears to, for several reasons:

Indirect infringement risk: If the model generating synthetic data was itself trained on copyrighted works without authorization, the synthetic data may retain a substantial part of that underlying real-world material and carry infringement liability downstream.⁴⁶

Training requirements: Producing useful synthetic data requires training a generator model on real-world data, creating the same copyright questions synthetic data purports to avoid.

Output similarity: Models trained on synthetic data may still generate outputs similar to copyrighted works from the original training data, particularly if the synthetic data preserves stylistic or structural patterns.

Practical value: Synthetic data quality degrades with each generation, limiting its utility for producing cutting-edge models.

Licensing Options: Building Compliant Training Datasets

As copyright litigation intensifies, licensing markets for training data have emerged rapidly. Companies have three primary licensing strategies:

Commercial Data Licensing

Major content providers now offer training data licenses:

News publishers:

• Providers: News Corp, Financial Times, Axel Springer, Le Monde
• License structure: Typically negotiated deals with major AI companies
• Cost range: Tens to hundreds of millions of dollars for comprehensive access
• Recent deals: OpenAI agreements with multiple publishers; Google’s deals with news organizations

Stock media platforms:

• Providers: Shutterstock, Getty Images, Adobe Stock
• License structure: API access to images, videos, and metadata with explicit AI training rights
• Cost range: Shutterstock’s chief executive told Bloomberg in June 2024 that the company’s AI licensing business produced $104 million in revenue in 2023; the company expects the segment to reach $250 million in sales by 2027⁴⁷
• Advantages: High-quality, commercially licensed content with clear rights

Publisher licensing:

• Providers: HarperCollins and other major publishers exploring AI licensing
• License structure: Author-approved licensing programs for book content
• Status: Emerging market with unclear pricing and terms
• Challenges: Complex rights (author vs. publisher ownership) and collective action problems

Code repositories:

• Providers: GitHub, GitLab potentially developing formal licensing programs
• Current status: Most code training relies on open-source licenses rather than commercial agreements
• Future development: Expect evolution toward opt-in commercial licensing as litigation clarifies risks

Creative Commons and Open Licensing

Creative Commons licenses provide graduated permission levels:

CC0 (Public Domain Dedication):

• Permission: No restrictions on any use, including commercial AI training
• Attribution: Not required but may be appreciated
• Best for: Maximum flexibility with zero compliance overhead
• Examples: Many government datasets, scientific databases, Common Corpus

CC BY (Attribution):

• Permission: Use, modification, and commercial application permitted
• Requirements: Provide attribution to original creator
• AI training considerations: Attribution requirement applies when publicly sharing works or adaptations; unclear whether trained models require attribution
• Examples: Many academic papers, educational resources, Wikipedia content

CC BY-SA (Attribution-ShareAlike):

• Permission: Use and commercial application permitted
• Requirements: Attribution plus derivatives must use same license
• AI training considerations: Unclear whether trained models constitute “derivatives” requiring ShareAlike; conservative approach treats models as subject to license
• Risk: Could require open-sourcing entire model if ShareAlike applies

CC BY-NC (Attribution-NonCommercial):

• Permission: Noncommercial use only
• Requirements: Attribution; all uses (including training) must be noncommercial
• AI training considerations: Commercial AI companies generally cannot use NC-licensed content, as both training and deployment involve commercial purposes
• Clarity: Most restrictive but clearest application—commercial AI training prohibited

CC BY-ND (Attribution-NoDerivatives):

• Permission: Redistribution permitted, modification prohibited
• Requirements: Attribution; no modifications or derivatives
• AI training considerations: Question whether training creates prohibited “derivatives”; conservative approach avoids these works
• Practical application: Often incompatible with AI training, which inherently transforms works

Important limitations: Creative Commons itself states that “using a more restrictive CC license in an effort to prevent AI training is not an effective approach” because copyright law may permit AI training regardless of license restrictions.⁴⁸ However, violating license terms creates contract-based liability even if copyright fair use might apply.

Public Domain Resources

Works published 1930 and earlier (pre-1931):

• Status: Definitively in U.S. public domain (the cutoff advances one year each January 1; 1930 works entered the public domain on January 1, 2026)
• Advantages: No copyright restrictions, no licensing costs
• Limitations: Historical works don’t reflect contemporary language, culture, or knowledge
• Examples: Classic literature, historical photographs, vintage artwork

U.S. government works:

• Status: Not subject to copyright under 17 U.S.C. § 105(a)⁴⁹
• Advantages: Vast repositories of technical, scientific, and administrative content
• Limitations: Excludes contractor-produced works; state government works may have copyright
• Examples: Federal agency reports, NASA images, legislative materials

Common Corpus and curated public domain datasets:

• Description: Approximately 2.27 trillion tokens across multiple languages, drawn from public domain and permissively licensed sources⁵⁰
• Advantages: Specifically curated for AI training with legal compliance
• Size: Significant but smaller than proprietary datasets
• Limitations: Represents only a portion of human knowledge; skews historical

Expiration and restoration: Copyright terms are complex. Works first published between 1931 and 1963 whose copyrights were not renewed in their twenty-eighth year fell into the public domain, but determining status requires research.⁵¹ Foreign works may have copyright restored under the Uruguay Round Agreements Act, creating unexpected liability for seemingly public domain content.⁵²

Cost Analysis: Licensing vs. Litigation Risk

Licensed Training Data Costs

Foundation model training (100B+ parameters):

• Commercial licensing costs: $50M to $500M+ depending on data volume, exclusivity, and content type
• Breakdown:
  • News content: $10M-100M+ per major publisher
  • Stock media: $5M-50M for image/video datasets
  • Code repositories: Variable, often relies on open-source rather than commercial licensing
  • Academic publishers: Emerging market, likely $10M-100M range

Specialized or fine-tuning datasets:

• Commercial licensing costs: $100K to $10M depending on domain specificity and volume
• Examples: Medical imaging data, financial news, technical documentation
• Structure: Often per-dataset pricing rather than comprehensive agreements

Open-source and public domain:

• Direct costs: Zero for most resources (CC BY, CC0, public domain)
• Compliance costs: $50K-500K for legal review, license tracking, attribution systems
• Limitations: Insufficient volume/quality for leading-edge models

Copyright Litigation Exposure

Statutory damages:

• Standard range: $750-$30,000 per infringed work⁵³
• Willful infringement: Up to $150,000 per work⁵⁴
• Innocent infringement: Minimum $200 per work (rarely applied to commercial entities)⁵⁵

Scale calculations:

If a foundation model is trained on 1 million copyrighted works (illustrative only):

• Minimum exposure (innocent): $200 per work × 1M works = $200 million
• Standard exposure: $750-$30,000 per work × 1M works = $750 million to $30 billion
• Willful infringement: $150,000 per work × 1M works = $150 billion

These figures are arithmetic illustrations of the statute’s per-work mechanics, not predictions; courts may question per-work aggregation across millions of dataset works.⁵⁶

Actual damages alternative:

Plaintiffs may elect actual damages (provable economic harm) plus defendant’s profits attributable to infringement instead of statutory damages. In practice, proving causation between specific training data and company profits presents challenges, making statutory damages more common in copyright litigation.

Anthropic settlement signal:

The roughly $3,000-per-work figure suggests a middle-ground resolution between the statutory minimum ($750) and the willful maximum ($150,000).⁵⁷ As a class settlement rather than a merits ruling, it sets no binding precedent, but it offers a real-world reference point for negotiated valuations where pirated source copies are at issue.

Defense costs:

• Motion to dismiss: $200K-$500K for comprehensive briefing and arguments
• Discovery through summary judgment: $2M-$10M including document production, depositions, expert witnesses
• Trial: $5M-$20M+ depending on complexity and duration
• Appeals: $1M-$3M per appellate level

Total defense costs for major litigation: $10M-$35M+ even if defendant ultimately prevails, not counting settlement amounts or judgments.

Risk-Adjusted Analysis

For companies training foundation models, licensing costs of $50M-$500M compare favorably to litigation exposure of hundreds of millions to billions in damages plus tens of millions in defense costs.

Break-even calculation:

Licensing becomes cost-effective if it eliminates:

• 10% probability of $500M judgment, or
• 50% probability of $100M judgment, or
• Certainty of $50M defense costs plus settlement

Most AI companies’ training practices face materially higher litigation risk than these thresholds, making proactive licensing economically rational for all but the smallest-scale operations—and the recent rulings sharpen one point in particular: clean provenance for source copies is among the cheapest forms of risk reduction available.

Strategic considerations beyond pure cost:

• Speed to market: Litigation delays product launches and fundraising
• Reputation risk: Copyright infringement allegations damage brand and customer relationships
• Enterprise sales: Many enterprise customers require vendor copyright indemnification
• Regulatory attention: Copyright violations attract regulatory scrutiny in the EU (AI Act) and U.S.⁵⁸

Compliance Framework: Building Defensible Training Data Programs

Data Sourcing Checklist

Implement systematic review for all training data sources:

• Document data provenance: Maintain records showing source, acquisition date, and legal basis for use
• Verify copyright status: Confirm works are public domain, licensed, or subject to fair use analysis
• Review licenses: For licensed content, verify license permits AI training and commercial use
• Implement filtering: Exclude high-risk categories (pirated works, recent creative content without authorization)
• Check opt-out lists: Respect robots.txt exclusions and explicit opt-out requests; providers of general-purpose AI models placed on the EU market must maintain a policy to identify and comply with rights reservations expressed by machine-readable means⁵⁹
• Assess fair use: For unlicensed copyrighted works, document fair use analysis for each category
• Third-party datasets: Investigate assembly methodology and copyright compliance for datasets obtained from third parties

Documentation Requirements

Courts and regulators increasingly require transparency about training data. Maintain comprehensive documentation:

Dataset documentation:

• Source and acquisition methodology
• Copyright status of included works
• License terms and compliance measures
• Opt-out mechanisms and responses
• Data processing and filtering applied

Fair use analysis:

• Written analysis applying four-factor test to each data category
• Business justification for using copyrighted works
• Market analysis of licensing alternatives
• Technical guardrails to prevent infringing outputs

EU AI Act compliance (for companies operating in the EU):

• Public summary of training data contents, using the AI Office’s mandatory template (required for general-purpose AI models from August 2, 2025)⁶⁰
• Documentation of copyright reservations respected
• Measures to remove illegal content
• Synthetic data generation details if applicable

Technical Safeguards

Implement technical measures to reduce infringement risk:

Input filtering:

• Block pirated or unauthorized sources at data collection stage
• Implement allowlists of authorized sources rather than blocklists
• Respect robots.txt and meta tag exclusions
• Honor DMCA takedown requests for specific works

Output filtering:

• Monitor generated outputs for substantial similarity to training data
• Implement guardrails preventing reproduction of copyrighted works
• Use perplexity filters or similarity detection to flag potential infringement
• Provide content filtering options for enterprise customers

Attribution systems:

• For CC BY licensed training data, develop attribution mechanisms
• Consider optional source attribution for generated content
• Implement transparency features showing training data categories

Audit capabilities:

• Maintain ability to identify whether specific works were included in training
• Implement logging for data usage and model training runs
• Develop response protocols for copyright holder inquiries

Organizational Practices

Cross-functional compliance team:

• Legal counsel for copyright analysis
• Engineering leadership for technical implementation
• Data acquisition team for sourcing decisions
• Compliance officer for ongoing monitoring

Board-level governance:

• Regular reporting on copyright compliance and litigation risk
• Executive approval for material training data decisions
• Budget allocation for licensing and compliance infrastructure

Vendor management:

• Due diligence on third-party data providers
• Contractual representations about copyright compliance
• Indemnification provisions for data provider breaches

Internal training:

• Educate engineers and data scientists on copyright compliance
• Establish clear escalation procedures for legal questions
• Foster culture of compliance rather than “move fast and break things”

Robots.txt and Opt-Out Mechanisms

Legal Status of Robots.txt

The robots.txt protocol is a voluntary technical standard without inherent legal force in the United States. Website owners place robots.txt files at domain roots to communicate preferences about automated crawling, but U.S. law does not explicitly require respecting these files.

However, practical and legal considerations favor compliance:

EU legal requirements: The Digital Single Market Directive’s text-and-data-mining exception applies only where the rightholder has not expressly reserved the use, including by machine-readable means, and the EU AI Act requires providers of general-purpose models to adopt a copyright policy that identifies and complies with those reservations, so ignoring a machine-readable opt-out takes training outside the exception and is inconsistent with the copyright policy the AI Act requires providers of general-purpose models to put in place and operate.⁶¹ Non-compliance risks regulatory enforcement and fines of up to 3 percent of annual worldwide turnover or EUR 15 million, whichever is higher, though the AI Office’s supervision and enforcement for general-purpose AI models begins on August 2, 2026.⁶¹

Terms of service: Violating website terms of service that incorporate robots.txt restrictions may support a breach-of-contract claim. A Computer Fraud and Abuse Act (CFAA) claim is narrower: the statute reaches one who “intentionally accesses a computer without authorization or exceeds authorized access, and thereby obtains … information from any protected computer,” and “exceeds authorized access” means “to access a computer with authorization and to use such access to obtain or alter information in the computer that the accesser is not entitled so to obtain or alter.”⁶² Whether ignoring robots.txt or terms of service on a publicly accessible site satisfies either element is a fact-specific question the statute’s text does not answer.

Fair use implications: Courts may view deliberately ignoring explicit opt-outs as evidence weighing against fair use, particularly on Factor 4 (market harm) when copyright holders attempt to preserve licensing markets.

Industry practice: Major AI companies including OpenAI and Google have published robots.txt guidance and claim to respect exclusions, making non-compliance a competitive and reputational disadvantage.

Implementing Opt-Out Respect

Technical implementation:

• Check robots.txt before scraping any domain
• Implement user-agent-specific rules (GPTBot, CCBot, etc.)
• Honor both robots.txt and meta tag exclusions
• Maintain allowlist of domains confirming permission

Common AI crawler user agents:

• GPTBot (OpenAI)
• Google-Extended (Google AI training)
• CCBot (Common Crawl)
• Bytespider (ByteDance)
• Anthropic-AI (Anthropic)

Retroactive compliance:

For training data already collected, some companies offer:

• Removal upon request from copyright holders
• Do-not-track mechanisms for future training runs
• Opt-out registries for artists, writers, and content creators

Documentation:

Maintain records of opt-out compliance for regulatory and litigation defense purposes.

Strategic Recommendations for AI Companies

For Early-Stage Startups

Immediate priorities:

1. Audit existing training data: Document sources and legal basis for all data currently used
2. Implement filtering: Remove highest-risk categories (pirated works, recent unlicensed creative content)
3. Shift to lower-risk sources: Prioritize public domain, permissive licenses, and commercial licensing for critical datasets
4. Document fair use analysis: Prepare written analysis for any remaining unlicensed copyrighted content
5. Budget for licensing: Allocate capital for data licensing in proportion to litigation risk

Funding stage considerations:

Pre-seed/seed stage: Use public domain and permissively licensed data (CC0, CC BY) exclusively if possible; consider pre-trained base models from vendors who handle copyright compliance

Series A/B stage: Begin commercial licensing relationships; implement comprehensive compliance framework; prepare for enterprise customer due diligence

Series C+ stage: Robust licensing portfolio; industry leadership in compliance practices; potential advocacy for favorable regulatory frameworks

For Enterprise AI Deployments

Companies deploying rather than developing models face different risks:

Vendor due diligence:

• Require vendors to disclose training data sources and copyright compliance
• Obtain contractual representations about copyright compliance
• Negotiate indemnification provisions for copyright claims
• Audit vendor compliance practices periodically

Fine-tuning and customization:

• Apply same compliance framework to any custom training data
• Ensure proprietary data licensing permits AI fine-tuning
• Document employee-generated content as company-owned

Open-source models:

• Investigate training data sources for open models
• Assess reputational and legal risks even if development liability lies with model creators
• Prefer models with documented compliant training data

For International Operations

EU-specific requirements:

The EU AI Act entered into force on August 1, 2024, and its obligations for general-purpose AI models—including the training-data-summary requirement—began applying on August 2, 2025. The Commission published the mandatory training-data-summary template on July 24, 2025,⁶³ and reissued the explanatory notice and template as a formal Communication on December 5, 2025. Models already on the market before August 2, 2025 have until August 2, 2027 to come into compliance.⁶⁴ Key obligations include:

• Public summary requirement: Publish a summary of training data contents, modalities, sizes, and sources, using the AI Office template
• Copyright compliance documentation: Demonstrate respect for copyright reservations and opt-outs
• Illegal content filtering: Document measures to remove illegal content from training data
• Synthetic data disclosure: If using synthetic data, explain generation methodology

Multi-jurisdictional strategy:

• Comply with strictest requirements (generally EU) for all markets
• Document compliance for each jurisdiction separately
• Monitor regulatory developments in key markets (UK, California, Canada)
• Consider jurisdiction-specific model variants if necessary

Looking Ahead: Regulatory and Legal Developments

Pending Legislation

AI-specific copyright legislation: Multiple proposals in Congress address AI training data rights, though passage timelines remain uncertain

State-level AI regulation: California, New York, and other states are considering AI bills that may address training data

Copyright Office Guidance Evolution

The May 2025 Copyright Office report is a pre-publication version and not the final word. Expect:

• A final version and revised guidance as case law develops
• Potential regulatory rulemaking on specific AI copyright issues
• Congressional requests for additional analysis as legislation advances

Industry Self-Regulation

AI industry associations are developing:

• Voluntary codes of practice for training data sourcing
• Licensing standards and best practices
• Opt-out registries and attribution systems

Companies participating in industry-led initiatives may receive favorable consideration from regulators and courts.

Litigation Timeline

Major cases proceeding through 2027:

• Andersen v. Stability AI: Jury trial September 20, 2027⁶⁵
• OpenAI MDL (S.D.N.Y.): Consolidated discovery underway, including the ordered production of roughly 20 million de-identified ChatGPT logs⁶⁶
• Thomson Reuters v. Ross: On interlocutory appeal to the Third Circuit⁶⁷
• GitHub Copilot: The Ninth Circuit affirmed the DMCA dismissal on September 16, 2026; the breach-of-license claims remain pending in the district court, which had stayed its proceedings for the appeal⁶⁸
• Anthropic settlement: Judgment entered July 20, 2026⁶⁹

Court decisions in these cases will continue to clarify fair use boundaries and establish damages reference points.

Key Takeaways

1. Copying is presumptively infringing, but the case law is split: Reproducing copyrighted works to build a training dataset implicates the reproduction right, yet 2025 rulings found the training step itself transformative—so the question is whether a fair use or other defense applies, not whether copying occurred

2. Provenance of source copies matters as much as the training: Bartz v. Anthropic treated model training as fair use while holding that the use of pirated copies was infringing; clean acquisition of source material is now a central risk variable⁷⁰

3. Factor 4 is record-dependent: Kadrey v. Meta turned on the plaintiffs’ failure to prove market dilution, with the court signaling a stronger record could change the outcome—market-harm evidence, not categorical rules, drives results⁷¹

4. Litigation costs are substantial: Even successful defense costs $10M-$35M+; resolutions range from roughly $3,000 per work to $1.5 billion in the aggregate (the Anthropic settlement)⁷²

5. Licensing markets are maturing: Commercial licensing options now exist across content types; costs range from millions to hundreds of millions but compare favorably to litigation exposure

6. Risk-based approach required: Assess copyright exposure by data source; prioritize lawfully obtained, licensed data for commercial applications; reserve fair use arguments for genuinely transformative uses

7. Technical safeguards are essential: Implement filtering, monitoring, and attribution systems to reduce infringement risk and demonstrate good faith

8. Documentation is critical: Courts and regulators demand transparency; maintain comprehensive records of data sourcing, legal analysis, and compliance measures

9. International compliance matters: The EU AI Act’s training-data transparency obligations for general-purpose models apply from August 2, 2025 (with a 2027 grace period for pre-existing models); U.S. companies with EU operations must comply⁷³

10. Monitor legal developments: Case law and regulatory guidance are evolving rapidly; quarterly compliance reviews are advisable

When to Seek Legal Counsel

Consult experienced AI copyright counsel when:

• Selecting training data sources for new models or significant training runs
• Evaluating fair use arguments for specific datasets
• Negotiating commercial licensing agreements with publishers or data vendors
• Responding to copyright holder inquiries or takedown requests
• Facing litigation or regulatory investigation
• Operating in multiple jurisdictions with varying copyright regimes
• Acquiring companies or technologies with uncertain training data provenance
• Developing enterprise sales where customers require copyright indemnification

---

Need AI Training Data Compliance Guidance?

Astraea Counsel advises AI companies on training data rights, copyright compliance, fair use analysis, and licensing strategies. We help AI startups navigate copyright risk while building compliant training datasets. Explore our AI & Emerging Tech services.

Related Resources

• California Frontier AI Law (SB 53) - State frontier-AI regulation framework
• Federal AI Regulation Landscape - Pending federal AI legislation
• Generative AI Copyright: Who Owns AI-Generated Content? - Ownership and human-authorship rules for AI outputs
• AI & Emerging Technology Practice - Comprehensive AI legal counsel
• Regulatory Compliance Services - Navigate AI compliance requirements
• Contact Us - Discuss your training data compliance needs

---

Disclaimer: This article provides general information only and does not constitute legal advice. Copyright law application to AI training involves complex, fact-specific analysis. Consult qualified legal counsel for advice on your specific situation.

---

Further Reading

• BakerHostetler, Case Tracker: Artificial Intelligence, Copyrights and Class Actions, available at https://www.bakerlaw.com/services/artificial-intelligence-ai/case-tracker-artificial-intelligence-copyrights-and-class-actions/

• Thorin Klosowski, No Robots(.txt): How to Ask ChatGPT and Google Bard to Not Use Your Website for Training, Electronic Frontier Foundation (Dec. 12, 2023), available at https://www.eff.org/deeplinks/2023/12/no-robotstxt-how-ask-chatgpt-and-google-bard-not-use-your-website-training

For dataset-level review and licensing negotiation, see our AI and autonomous agent counsel .

---

Footnotes

1. Bartz v. Anthropic PBC, 787 F. Supp. 3d 1007, 1025, 1033-34 (N.D. Cal. 2025). PDF ↩

2. Order Granting Final Approval of Class Action Settlement, Granting in Part Motion for Attorney’s Fees, and Judgment, Bartz v. Anthropic PBC, No. 3:24-cv-05417-AMO, Dkt. No. 680, at 4 (N.D. Cal. July 20, 2026) (“The non-reversionary Settlement Fund of $1.5 billion”), available at https://storage.courtlistener.com/recap/gov.uscourts.cand.434709/gov.uscourts.cand.434709.680.0_5.pdf. PDF ↩

3. U.S. Copyright Office, Copyright and Artificial Intelligence, Part 3: Generative AI Training 26 (pre-publication version, May 2025) (“The steps required to produce a training dataset containing copyrighted works clearly implicate the right of reproduction.”), available at https://www.copyright.gov/ai/Copyright-and-Artificial-Intelligence-Part-3-Generative-AI-Training-Report-Pre-Publication-Version.pdf. PDF ↩

4. 17 U.S.C. § 106(1) (2026). PDF ↩

5. 17 U.S.C. § 107 (2026). PDF ↩

6. Bartz, 787 F. Supp. 3d at 1019. PDF ↩

7. Copyright Office Part 3 Report, supra note 3, at 46 (footnote omitted) (“[T]raining a model is most transformative when the purpose is to deploy it for research, or in a closed system that constrains it to a non-substitutive task.”; “On the other end of the spectrum is training a model to generate outputs that are substantially similar to copyrighted works in the dataset.”; “Where a model is trained on specific types of works in order to produce content that shares the purpose of appealing to a particular audience, that use is, at best, modestly transformative.”). PDF ↩ ↩²

8. Copyright Office Part 3 Report, supra note 3, at 47—48. PDF ↩

9. 17 U.S.C. § 107 (2026) (“The fact that a work is unpublished shall not itself bar a finding of fair use if such finding is made upon consideration of all the above factors.”). PDF ↩

10. Copyright Office Part 3 Report, supra note 3, at 57 (“research supports commenters’ assertions that internet-scale pre-training data, including large amounts of entire works, may be necessary to achieve the performance of current-generation models”). PDF ↩ ↩²

11. Copyright Office Part 3 Report, supra note 3, at 61—71 (market-harm analysis, including lost licensing opportunities; “Where licensing markets are available to meet AI training needs, unlicensed uses will be disfavored under the fourth factor. But if barriers to licensing prove insurmountable for parties’ uses of some types of works, there will be no functioning market to harm and the fourth factor may favor fair use.”). PDF ↩ ↩²

12. Kadrey v. Meta Platforms, Inc., 788 F. Supp. 3d 1026, 1060 (N.D. Cal. 2025). PDF ↩

13. Bartz, 787 F. Supp. 3d at 1019, 1021, 1025, 1033-34. PDF ↩

14. Kadrey, 788 F. Supp. 3d at 1034, 1060. PDF ↩

15. Thomson Reuters Enter. Ctr. GmbH v. Ross Intel. Inc., 694 F. Supp. 3d 467, 487 (D. Del. 2023) (“So a jury must decide the fourth factor—and the ultimate conclusion on fair use.”), revised by 765 F. Supp. 3d 382 (D. Del. 2025). PDF ↩

16. Thomson Reuters, 765 F. Supp. 3d at 397-98. PDF ↩

17. Thomson Reuters, 765 F. Supp. 3d at 400 (quoting Harper & Row, Publishers, Inc. v. Nation Enters., 471 U.S. 539, 566 (1985)). PDF ↩

18. Thomson Reuters, 765 F. Supp. 3d at 401. PDF ↩

19. Thomson Reuters Enter. Ctr. GmbH v. Ross Intel. Inc., 2025 U.S. App. LEXIS 35326 (3d Cir. June 17, 2025) (granting the petition for interlocutory appeal); appeal filed June 24, 2025, per the subsequent history to 765 F. Supp. 3d 382. PDF ↩

20. Andersen v. Stability AI Ltd., 744 F. Supp. 3d 956, 969, 986 (N.D. Cal. 2024). PDF ↩

21. Docket, Andersen v. Stability AI Ltd., No. 3:23-cv-00201-WHO (N.D. Cal.), https://www.courtlistener.com/docket/66732129/andersen-v-stability-ai-ltd/ (Dkt. 597, Order Regarding Case Schedule (June 15, 2026) (jury trial reset for September 20, 2027; pretrial conference August 23, 2027)). ↩

22. Andersen, 744 F. Supp. 3d at 969. PDF ↩

23. Andersen, 744 F. Supp. 3d at 974—75, 982. PDF ↩

24. Complaint ¶ 9, New York Times Co. v. Microsoft Corp., No. 1:23-cv-11195, Dkt. No. 1, at 4 (S.D.N.Y. Dec. 27, 2023) (“This action seeks to hold them responsible for the billions of dollars in statutory and actual damages that they owe for the unlawful copying and use of The Times’s uniquely valuable works.”), available at https://storage.courtlistener.com/recap/gov.uscourts.nysd.612697/gov.uscourts.nysd.612697.1.0.pdf. PDF ↩

25. Opinion & Order, New York Times Co. v. Microsoft Corp., No. 1:23-cv-11195-SHS-OTW, Dkt. No. 344, at 2, 6 (S.D.N.Y. Nov. 22, 2024) (Wang, M.J.) (“Because Defendant has not demonstrated the relevance of the information sought, their motion to compel is DENIED.”), available at https://storage.courtlistener.com/recap/gov.uscourts.nysd.612697/gov.uscourts.nysd.612697.344.0.pdf; objections overruled, Order, Dkt. No. 592, at 1, 12 (S.D.N.Y. May 30, 2025) (Stein, J.) (denying defendants’ Rule 72(a) motions to set aside because the orders “are neither clearly erroneous nor contrary to law”), available at https://storage.courtlistener.com/recap/gov.uscourts.nysd.612697/gov.uscourts.nysd.612697.592.0.pdf. PDF PDF ↩

26. In re OpenAI, Inc., 776 F. Supp. 3d 1352, 1356 (J.P.M.L. 2025). PDF ↩

27. Davis v. OpenAI (In re OpenAI, Inc. Copyright Infringement Litig.), 2025 U.S. Dist. LEXIS 259061, at *76 (S.D.N.Y. Dec. 15, 2025). PDF ↩

28. In re OpenAI, Inc., 2025 U.S. Dist. LEXIS 222099, at *49-50 (S.D.N.Y. Nov. 7, 2025), aff’d and objection overruled, 2026 U.S. Dist. LEXIS 1986 (S.D.N.Y. Jan. 5, 2026). PDF ↩

29. In re OpenAI, Inc., 2025 U.S. Dist. LEXIS 252574, at *71-72 (S.D.N.Y. Dec. 2, 2025) (Wang, M.J.). PDF ↩

30. Doe v. GitHub, Inc., 672 F. Supp. 3d 837, 859—60, 862 (N.D. Cal. 2023). PDF ↩

31. 17 U.S.C. § 1202(b) (2026). PDF ↩

32. Doe v. GitHub, Inc., 2024 U.S. Dist. LEXIS 11068, at *25-27 (N.D. Cal. Jan. 3, 2024) (dismissing the Section 1202(b)(1) and (b)(3) claims with leave to amend for failure to plead identicality), reconsideration denied, 2024 U.S. Dist. LEXIS 68395 (N.D. Cal. Apr. 15, 2024). The order certified for interlocutory appeal was the court’s subsequent June 24, 2024 dismissal of the repleaded claims. PDF ↩

33. Doe v. GitHub, Inc., 2024 U.S. Dist. LEXIS 175951, at *9 (N.D. Cal. Sept. 27, 2024) (certifying the Section 1202(b) dismissal for interlocutory appeal and staying trial court proceedings); Order of USCA, Doe v. GitHub, Inc., No. 4:22-cv-06823-JST (N.D. Cal. Dec. 19, 2024), Dkt. 285 (“The petition for permission to appeal is granted.”), https://www.courtlistener.com/docket/65669506/doe-1-v-github-inc/; Docket, Doe v. GitHub, Inc., No. 24-7700 (9th Cir. docketed Dec. 23, 2024), https://www.courtlistener.com/docket/69495342/doe-et-al-v-github-inc-et-al/ (appeal opened on the petition granted in No. 24-6136). PDF PDF ↩

34. Doe v. GitHub, Inc., No. 24-7700, slip op. at 6, 9—11, 13, 15, 17—18 (9th Cir. Sept. 16, 2026) (Copilot and Codex “do not ‘remove or alter’ copyright management information (CMI) from a copy of an existing protected work but instead create new works that never contained that information”; “‘identicality’ is something of a misnomer because the DMCA does not require literal identicality between the plaintiff’s work and the allegedly infringing work”; “best understood as a gloss on the statutory terms ‘remove,’ ‘alter,’ and ‘copies’ rather than an independent (and atextual) element of a section 1202(b) claim”; “plaintiffs’ own allegations about Copilot show that it is best understood as learning from existing works and then creating new works based on that learning process, not as making copies of existing works”; “We express no view on whether that similarity would allow plaintiffs to assert a claim for copyright infringement.”; “The court denied a motion to dismiss the contract claims, which remain pending.”). PDF ↩

35. Bartz v. Anthropic PBC, 2025 U.S. Dist. LEXIS 205531, at *14-16 (N.D. Cal. Oct. 17, 2025) (per-work award of about $3,000, less costs and fees, across 482,460 works finally identified); Order Granting Final Approval of Class Action Settlement, Granting in Part Motion for Attorney’s Fees, and Judgment, Bartz v. Anthropic PBC, No. 3:24-cv-05417-AMO, Dkt. No. 680, at 4 (N.D. Cal. July 20, 2026) (440,490 of the 482,460 Works on the Works List claimed as of April 16, 2026). PDF PDF ↩

36. Bartz, 2025 U.S. Dist. LEXIS 205531, at *7-8, *32. PDF ↩

37. Order Granting Final Approval of Class Action Settlement, Granting in Part Motion for Attorney’s Fees, and Judgment, Bartz v. Anthropic PBC, No. 3:24-cv-05417-AMO, Dkt. No. 680, at 1, 23 (N.D. Cal. July 20, 2026) (“The motions were heard before this Court on May 14, 2026.”), available at https://storage.courtlistener.com/recap/gov.uscourts.cand.434709/gov.uscourts.cand.434709.680.0_5.pdf. PDF PDF ↩

38. Bartz, 2025 U.S. Dist. LEXIS 205531, at *16-17 (fee request capped at 25 percent before costs and expenses); Order Granting Final Approval of Class Action Settlement, Granting in Part Motion for Attorney’s Fees, and Judgment, Bartz v. Anthropic PBC, No. 3:24-cv-05417-AMO, Dkt. No. 680, at 4—5 (quoting Settlement Agreement ¶ 8.1: “Class Counsel shall be paid the Fee Award in equal installments proportionate to Anthropic’s payments to the Settlement Fund, as described above, provided that no Fee Award payment shall be paid to Class Counsel from the Settlement Fund until after the Final Approval.”). PDF PDF ↩

39. Order Granting Final Approval of Class Action Settlement, Granting in Part Motion for Attorney’s Fees, and Judgment, Bartz v. Anthropic PBC, No. 3:24-cv-05417-AMO, Dkt. No. 680, at 4, 14, 18, 22 (N.D. Cal. July 20, 2026) (plaintiffs moved for 12.5 percent, $187,500,000, of the $1.5 billion Settlement Fund; the court found “a multiplier of 3.75, for a total fee award of $101,561,111 is reasonable” and granted fees “in the amount of $101,561,111, nearly 6.8% percent of the non-reversionary Settlement Fund”; “As of April 16, 2026, 91.3 percent of the Works List had been claimed”). PDF ↩

40. 17 U.S.C. § 504(c)(1) (2026); Bartz, 2025 U.S. Dist. LEXIS 205531, at *15 (“four times the statutory minimum for ordinary infringement … of $750”). PDF PDF ↩

41. Anthropic PBC’s Reply in Support of Motion for Stay, Bartz v. Anthropic PBC, No. 3:24-cv-05417-WHA, Dkt. No. 278, at 3, 5 (N.D. Cal. July 30, 2025) (a class trial would expose “Anthropic alone to potentially hundreds of billions of dollars of liability”; “A damages award of that magnitude—potentially hundreds of billions of dollars—would be devastating for Anthropic”), available at https://storage.courtlistener.com/recap/gov.uscourts.cand.434709/gov.uscourts.cand.434709.278.0.pdf. PDF ↩

42. 17 U.S.C. § 504(c)(2) (2026). PDF ↩

43. Bartz, 787 F. Supp. 3d at 1033-34; Final Approval Order and Judgment, supra note 2, at 4. PDF PDF ↩

44. Thomson Reuters, 765 F. Supp. 3d at 390—91, 395—96, 401. PDF ↩

45. Doe v. GitHub, 672 F. Supp. 3d at 860; Doe v. GitHub, Inc., No. 24-7700, slip op. at 9, 18 (9th Cir. Sept. 16, 2026). PDF PDF ↩

46. Gareth Kristensen et al., Training AI Models on Synthetic Data: No Silver Bullet for IP Infringement Risk in the Context of Training AI Systems (Part 2 of 4), Cleary AI & Tech. Insights (Dec. 19, 2023), https://www.clearyiptechinsights.com/2023/12/training-ai-models-on-synthetic-data-no-silver-bullet-for-ip-infringement-risk-in-the-context-of-training-ai-systems-part-2-of-4/ (“Synthetic data itself may also not be entirely shielded from IP infringement claims to the extent it retains in substantial part the initial real-world data on which the generating AI system was trained.”). ↩

47. Brody Ford, Shutterstock’s AI-Licensing Business Generated $104 Million Last Year, Bloomberg (June 4, 2024), syndicated at https://finance.yahoo.com/news/shutterstock-ai-licensing-business-generated-120000890.html (“The licensing business with AI companies produced $104 million in revenue last year”; “By 2027, Shutterstock expects the segment to produce sales of $250 million”). ↩

48. Sarah Hinchliff Pearson, Understanding CC Licenses and AI Training: A Legal Primer, Creative Commons (May 15, 2025), https://creativecommons.org/2025/05/15/understanding-cc-licenses-and-ai-training-a-legal-primer/ (“AI training is often permitted by copyright… . This also means that using a more restrictive CC license in an effort to prevent AI training is not an effective approach.”). ↩

49. 17 U.S.C. § 105(a) (2026). PDF ↩

50. Pleias, Common Corpus (dataset card), https://huggingface.co/datasets/PleIAs/common_corpus (“Common Corpus is the largest open licensed text dataset, comprising 2.27 trillion tokens (2,267,302,720,836 tokens).”; “All data in Common Corpus are either uncopyrighted or freely licensed and may be used for both commercial and non-commercial purposes.”) (live dataset card; figures as of Sept. 18, 2026). ↩

51. 17 U.S.C. § 304(a) (2024); id. § 304 note (Amendments, 1992) (for copyrights secured before January 1, 1964, § 304(a) as in effect before June 26, 1992 governs, under which “in default of the registration of such application for renewal and extension, the copyright in any work shall terminate at the expiration of twenty-eight years from the date copyright was originally secured”); id. § 304 note (Effective Date of 1992 Amendment) (automatic renewal “applicable only to copyrights secured between January 1, 1964, and December 31, 1977”); id. § 304(b) (a copyright in its renewal term when the Sonny Bono Copyright Term Extension Act took effect “shall have a copyright term of 95 years from the date copyright was originally secured”). PDF ↩

52. 17 U.S.C. § 104A (2026). PDF ↩

53. 17 U.S.C. § 504(c)(1) (2026). PDF ↩

54. 17 U.S.C. § 504(c)(2) (2026). PDF ↩

55. 17 U.S.C. § 504(c)(2) (2026). PDF ↩

56. 17 U.S.C. § 504(c) (2026). PDF ↩

57. Bartz, 2025 U.S. Dist. LEXIS 205531, at *14-15; 17 U.S.C. § 504(c) (2026). PDF PDF ↩

58. Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence (AI Act), 2024 O.J. (L) 1689, http://data.europa.eu/eli/reg/2024/1689/oj. PDF ↩

59. AI Act, supra note 55, art. 53(1)(c) (providers of general-purpose AI models shall “put in place a policy to comply with Union law on copyright and related rights, and in particular to identify and comply with, including through state-of-the-art technologies, a reservation of rights expressed pursuant to Article 4(3) of Directive (EU) 2019/790”). PDF ↩

60. AI Act, supra note 55, arts. 53(1)(d) (providers shall “draw up and make publicly available a sufficiently detailed summary about the content used for training of the general-purpose AI model, according to a template provided by the AI Office”), 113(b); see also id. art. 111(3) (providers of general-purpose AI models placed on the market before 2 August 2025 to comply by 2 August 2027); European Commission, Explanatory Notice and Template for the Public Summary of Training Content for General-Purpose AI Models, C(2025) 8311 final, ¶ 33 (Dec. 5, 2025), https://ec.europa.eu/newsroom/dae/redirection/document/118480. PDF PDF ↩

61. AI Act, supra note 55, art. 53(1)(c); Directive (EU) 2019/790 of the European Parliament and of the Council of 17 April 2019 on copyright and related rights in the Digital Single Market, art. 4(3), 2019 O.J. (L 130) 92 (the text-and-data-mining exception applies “on condition that the use of works and other subject matter referred to in that paragraph has not been expressly reserved by their rightholders in an appropriate manner, such as machine-readable means in the case of content made publicly available online”), http://data.europa.eu/eli/dir/2019/790/oj; AI Act, supra note 55, arts. 101(1) (fines on providers of general-purpose AI models “not exceeding 3 % of their annual total worldwide turnover in the preceding financial year or EUR 15 000 000, whichever is higher”), 113(b) (Chapter V and Chapter XII apply from 2 August 2025, “with the exception of Article 101”); C(2025) 8311 final, supra note 57, ¶ 33 (“The supervision and enforcement by the AI Office for compliance with the rules for general-purpose AI models will start as of 2 August 2026.”). PDF PDF PDF ↩ ↩²

62. 18 U.S.C. § 1030(a)(2)(C), (e)(6) (2026). PDF ↩

63. European Commission, Explanatory Notice and Template for the Public Summary of Training Content for General-Purpose AI Models, Shaping Europe’s Digital Future (July 24, 2025), https://digital-strategy.ec.europa.eu/en/library/explanatory-notice-and-template-public-summary-training-content-general-purpose-ai-models (library page dated “Publication 24 July 2025”). ↩

64. AI Act, supra note 55, arts. 53(1)(d), 111(3), 113(b) (obligations for providers of general-purpose AI models apply from 2 August 2025; “Providers of general-purpose AI models that have been placed on the market before 2 August 2025 shall take the necessary steps in order to comply with the obligations laid down in this Regulation by 2 August 2027”); European Commission, C(2025) 8311 final, supra note 57, ¶ 33 (“The obligation for making the Summaries publicly available becomes applicable as of 2 August 2025. For models placed on the market before 2 August 2025, providers should take the necessary steps to make the corresponding Summary publicly available no later than 2 August 2027.”). PDF PDF ↩

65. Andersen, 744 F. Supp. 3d 956; Docket, supra note 21. PDF ↩

66. In re OpenAI, Inc., 2025 U.S. Dist. LEXIS 222099, at *49-50 (S.D.N.Y. Nov. 7, 2025), aff’d and objection overruled, 2026 U.S. Dist. LEXIS 1986 (S.D.N.Y. Jan. 5, 2026); In re OpenAI, Inc., 2025 U.S. Dist. LEXIS 252574, at *71-72 (S.D.N.Y. Dec. 2, 2025). PDF PDF ↩

67. Thomson Reuters, 765 F. Supp. 3d 382; see supra note 19. PDF ↩

68. Doe v. GitHub, Inc., No. 24-7700, slip op. at 9, 18 (9th Cir. Sept. 16, 2026) (“The court denied a motion to dismiss the contract claims, which remain pending.”; “AFFIRMED.”); Doe v. GitHub, Inc., 2024 U.S. Dist. LEXIS 175951, at *9 (N.D. Cal. Sept. 27, 2024) (staying trial court proceedings pending the appeal). PDF PDF ↩

69. Final Approval Order and Judgment, supra note 2, at 23; see also Bartz, 787 F. Supp. 3d 1007 (subsequent history) (judgment entered July 20, 2026). PDF PDF ↩

70. Bartz, 787 F. Supp. 3d at 1019, 1033-34. PDF ↩

71. Kadrey, 788 F. Supp. 3d at 1060. PDF ↩

72. Bartz, 2025 U.S. Dist. LEXIS 205531, at *14-15; Order Granting Final Approval of Class Action Settlement, Granting in Part Motion for Attorney’s Fees, and Judgment, Bartz v. Anthropic PBC, No. 3:24-cv-05417-AMO, Dkt. No. 680, at 4 (N.D. Cal. July 20, 2026) ($1.5 billion non-reversionary Settlement Fund). PDF PDF ↩

73. AI Act, supra note 55, arts. 53(1)(c)—(d), 111(3), 113(b); C(2025) 8311 final, supra note 57, ¶ 33. PDF PDF ↩