← Back to Hub
⚖️ AI Copyright Cases 2022–2025

NYT vs OpenAI: $10–100B
Training Data Liability

The New York Times. Getty Images. Authors Guild. The music industry. They're all suing AI companies for training on their content without permission. The outcome will reshape the entire industry.

$10B+ Estimated low-end liability if NYT wins
$100B High-end liability if copyright theory applies broadly
27+ Active copyright lawsuits against AI companies (2024)
2026 Expected first major ruling (NYT v. OpenAI)

Choose your depth. The data doesn't change — just the explanation.

AI companies trained their systems by reading billions of articles, books, and photos that other people created. They did this without asking or paying. Now the creators are suing. The New York Times is the biggest case — if they win, it could cost AI companies billions of dollars and make AI much harder to build in the future.
Training AI on copyrighted content raises a fundamental legal question: does copying data to train a model constitute copyright infringement? AI companies argue "fair use" (transformative purpose, no market substitution). Plaintiffs argue systematic copying at scale destroys the market for the original work. The NYT suit specifically showed GPT-4 reproducing verbatim NYT articles — which dramatically weakens the "no market substitution" fair use defense. If courts rule against AI companies, they face retroactive liability for all past training + must either license or retrain.
Key legal framework: 17 U.S.C. § 107 (fair use factors). Authors Guild v. Google (2d Cir. 2015): book scanning = fair use (transformative, search snippets). But AI differs: outputs can substitute for originals directly. NYT complaint: model directly reproduced 100+ full NYT articles verbatim. Getty Images v. Stability AI: UK + US cases, UK includes sui generis database rights. Music cases (Concord v. Anthropic): lyrics reproduction in RAG. The § 1201 DMCA angle: circumventing opt-out mechanisms = separate violation. Statutory damages: $750-$150,000 per work infringed × millions of works = astronomical potential liability.
NYT v. Microsoft/OpenAI: SDNY 1:23-cv-11195. Getty v. Stability AI: D.Del. 1:23-cv-00135, UK High Court Ch 2023-000007. Authors Guild v. OpenAI: SDNY 1:23-cv-08292. Concord v. Anthropic (lyrics): M.D. Tenn. 3:23-cv-01092. Key precedents: Authors Guild v. Google (804 F.3d 202); Campbell v. Acuff-Rose (510 U.S. 569). Track via CourtListener (courtlistener.com) and Lex Machina. Statutory damages calculator: works × $150K (willful) = max exposure. Likely settlement range: $1-5B for OpenAI (precedent: music industry vs. Napster).

Active Cases Timeline

27+ active lawsuits form a legal ecosystem that could determine whether the entire AI industry's training data practices are legal.

Estimated Liability by Case ($B)

Statutory damages if plaintiffs prevail. Actual settlements will be lower. For context only.

Dec 2023
New York Times
NYT v. Microsoft / OpenAI — SDNY
Alleges GPT-4 reproduced 100+ full articles verbatim. Seeks damages up to $150K/work. OpenAI's training data included millions of NYT articles. Landmark case.
Pending — SDNY
Jan 2023
Getty Images
Getty Images v. Stability AI — US & UK
Getty's watermark appeared in Stable Diffusion outputs. Seeks $1.8B+. UK High Court case ongoing separately under UK copyright law. Image training data practices.
Pending — D. Delaware
Sep 2023
Authors Guild + 17 authors
Authors Guild v. OpenAI — SDNY
John Grisham, George R.R. Martin, Jodi Picoult among plaintiffs. Books3 training dataset contained 196,640 books. Class action potential.
Pending — SDNY
Oct 2023
Concord Music Group
Concord v. Anthropic — M.D. Tennessee
Claude reproduces song lyrics verbatim on request. Music publishers allege systematic infringement. First major music industry AI suit.
Pending — M.D. Tenn.
2024
Alden Global / newspapers
Multiple newspaper publishers v. OpenAI
8 major newspaper publishers joined. Chicago Tribune, Denver Post, etc. Part of coordinated IP strategy. Same legal theory as NYT.
Pending — SDNY (consolidated)

Case Count Growth (Active AI Copyright Suits)

Cumulative active cases by month, 2022–2024.

Plaintiff Types (% of Active Suits)

Who is suing AI companies over training data.

The Nuclear Option

If courts apply statutory copyright damages broadly — $750 to $150,000 per work infringed — and AI training datasets contain hundreds of millions of copyrighted works, theoretical liability exceeds the GDP of small countries. This is why the industry is desperate to establish fair use precedent before cases reach verdict.

Sources

• NYT v. Microsoft/OpenAI. SDNY 1:23-cv-11195. Filed December 27, 2023.

• Getty Images v. Stability AI. D.Del. 1:23-cv-00135; UK High Court Ch 2023-000007.

• Authors Guild v. OpenAI. SDNY 1:23-cv-08292.

• Concord Music Group v. Anthropic. M.D. Tenn. 3:23-cv-01092.

• CourtListener. courtlistener.com — Track AI copyright dockets.