AI Copyright Lawsuits Outside the US: EU, UK & India

AI copyright lawsuits are not just an American story. While Ban the Bots' AI lawsuit tracker covers dozens of active US federal cases against OpenAI, Meta, Anthropic, and others, courts and regulators in the European Union, United Kingdom, and India are running their own — often stricter — legal tests of whether training AI on copyrighted material is lawful. Some have already ruled against AI companies in ways no US court has. This page tracks the real, verifiable cases and regulatory actions outside the US, and is honest about where search-demand hype (Minimax, Sora) outpaces what's actually been filed in court.

• Why AI copyright fights look different outside the US
• European Union: GEMA v. OpenAI and the AI Act's copyright rules
• United Kingdom: Getty Images v. Stability AI
• India: ANI Media v. OpenAI
• Australia: no lawsuit yet, but real legal exposure
• What about Minimax Hailuo and Sora?
• FAQ

Why AI copyright fights look different outside the US

Every major US AI copyright case turns on fair use — a flexible, four-factor doctrine that lets a court excuse copying if it's "transformative" enough. Judge Chhabria's 2025 ruling for Meta and Judge Alsup's ruling for Anthropic both leaned on that flexibility (see the underlying cases on /ai-lawsuits/). Most other countries don't have anything like it. The EU and UK instead have a narrow "text and data mining" exception that arguably covers only the analysis phase of training, not a model memorizing and reproducing a work. India and Australia rely on enumerated "fair dealing" purposes — research, review, news reporting — that commercial AI training doesn't obviously fit. The same practice a US judge might excuse as "transformative" can be a straightforward infringement claim elsewhere. That's exactly what happened in Germany.

European Union: GEMA v. OpenAI and the AI Act's copyright rules

On November 11, 2025, the Munich Regional Court I (case no. 42 O 14139/24) issued what several law firms call Europe's first ruling holding an AI developer directly liable for training-data copyright infringement. German royalty society GEMA sued OpenAI over nine well-known German song lyrics — including Herbert Grönemeyer's "Männer" — that ChatGPT would reproduce almost verbatim from simple prompts.

The court found this "memorization" and "regurgitation" is a reproduction under EU and German copyright law, falling outside the EU's text-and-data-mining exception, which it said covers only the initial analytical phase of training, not a model storing and later outputting protected expression. The court largely sided with GEMA on injunctive relief, disclosure, and damages. OpenAI has said it will appeal, so the ruling isn't final, but it's already reshaping how European rightsholders frame claims.

Separately, Italy's data protection authority fined OpenAI €15 million in December 2024 over ChatGPT's handling of personal training data — the first GDPR enforcement action against a generative AI company, though a Rome court annulled the fine in March 2026 (full details on /ai-lawsuits/). The EU is also regulating proactively: since August 2025, the AI Act requires "general-purpose AI" providers to maintain a copyright policy and publish a public summary of their training data, per an EU AI Office template — a binding transparency regime no US law currently matches.

United Kingdom: Getty Images v. Stability AI

The UK's marquee case is Getty Images v. Stability AI [2025] EWHC 2863 (Ch), decided on November 4, 2025 — the first UK ruling to squarely address generative AI and copyright. Getty had accused Stability AI of training its Stable Diffusion model on millions of Getty photos without a license.

The result was mostly a loss for Getty on copyright. Unable to show the actual training took place in the UK, it dropped its primary copyright and database-right claims before closing arguments. On the narrower claim that remained — whether importing the trained model into the UK was "secondary infringement" — the court ruled that a model's weights are statistically trained parameters, not a stored "copy" of any photograph, so no infringement occurred. Getty did win a limited trademark claim over its watermark appearing in some outputs, and has permission to appeal the copyright question to the Court of Appeal, so this isn't the final word in the UK.

For context on the artist-side version of this fight, see Ban the Bots' AI art theft and copyright opt-outs explainer.

India: ANI Media v. OpenAI

India's first major AI copyright case is playing out at the Delhi High Court. Asian News International (ANI), a major Indian news wire, sued OpenAI in November 2024, arguing ChatGPT was trained on its news content without a license and had fabricated quotes wrongly attributed to ANI. ANI seeks roughly ₹2 crore (about $240,000) plus an injunction.

OpenAI argues Indian courts lack jurisdiction since its servers and training operations sit outside India, and that training qualifies as "fair dealing" for research under Section 52 of the Copyright Act, 1957. The Digital News Publishers Association and other Indian outlets have intervened, making this a bellwether for the country's news industry. After roughly 32 hearings between November 2024 and March 2026, Justice Amit Bansal reserved his order. As of this writing (July 2026), no ruling has issued — it's one of the most closely watched pending AI copyright decisions worldwide, since India has no fair-use doctrine to fall back on the way US defendants do.

Australia: no lawsuit yet, but real legal exposure

Unlike the US, EU, UK, and India, no AI-copyright-training lawsuit has yet been filed in an Australian court — worth saying plainly rather than implying otherwise. Australian artists have publicly accused tools like Lensa AI of training on their work without consent, and law firms there have written extensively about the risk, but as of mid-2026 that anger hasn't turned into a filed case.

It may only be a matter of time: Australian copyright law, like India's, uses narrow fair-dealing exceptions (research, criticism, parody, news reporting) rather than the open-ended US fair-use test, and commercial AI training doesn't cleanly fit any of them. Commentators there point to Getty v. Stability AI and the US Anthropic/Meta settlements as previews of arguments an Australian case would raise, since the Copyright Act 1968 offers AI developers a thinner defense than US law does.

What about Minimax Hailuo and Sora?

Both show up constantly in search demand, so it's worth being precise about what's actually filed.

Minimax Hailuo: Disney, Universal, and Warner Bros. Discovery did sue Shanghai-based MiniMax over its Hailuo video generator in September 2025, alleging it trained on their films and could output near-identical clips of characters like Darth Vader and the Minions. But this isn't a foreign-court case — it was filed in the U.S. District Court for the Central District of California, the same venue as most cases on /ai-lawsuits/. A judge denied MiniMax's motion to dismiss in 2026; the case is headed to discovery — real and significant, just not "outside the US" litigation despite the Chinese defendant.

Sora: OpenAI's video model drew huge controversy after its September 2025 relaunch, including a public letter from the Motion Picture Association demanding OpenAI stop copyrighted characters appearing in outputs, and a policy reversal to an opt-in system for rightsholders. We could not verify a distinct, filed lawsuit targeting Sora's outputs specifically, separate from the broader suits OpenAI already faces (tracked on /ai-lawsuits/). OpenAI shut the standalone app down in April 2026. If a Sora-specific suit is filed, we'll add it — not invent one to match search traffic.

FAQ

Has any country ruled against an AI company on copyright?

Yes. Germany's Munich Regional Court ruled against OpenAI in GEMA v. OpenAI (November 2025), finding ChatGPT's reproduction of memorized song lyrics is infringement not covered by the EU's text-and-data-mining exception. OpenAI is appealing.

Did Getty Images win its case against Stability AI?

Mostly no. The UK High Court rejected Getty's core copyright and database-right claims in November 2025, though it won a narrow trademark claim, and is appealing the copyright ruling.

Is the India ANI v. OpenAI case decided yet?

Not as of July 2026. The Delhi High Court has reserved judgment after more than 30 hearings, in one of the first tests of AI training under Indian copyright law.

Is there an AI copyright lawsuit in Australia?

Not yet, as of mid-2026, though commentators expect one given Australia's narrower fair-dealing exceptions.

Why do EU and UK rules treat AI training differently than the US?

The US uses "fair use," an open-ended test AI companies argue is "transformative." The EU, UK, India, and Australia use narrow, listed exceptions that don't obviously cover commercial AI training — part of why GEMA won in Germany when authors lost key US rounds.

The pattern to watch

The US isn't necessarily where AI copyright law will end up settled first. Germany has already ruled against OpenAI once. The UK's Getty case is headed to appeal on the core copyright question. India's Delhi High Court could rule any day on a case with no US-style fair-use safety valve. And the EU's AI Act now forces training-data transparency no AI company has to disclose in the US. For the parallel US docket — Anthropic, Meta, OpenAI, Stability, and dozens more — see the full AI lawsuit tracker, and for how this affects working artists, see AI art theft and copyright opt-outs.