Training Was Fair Use. Pirated Copies Were Not. The Copyright Map So Far.

Around 130 AI copyright lawsuits are being tracked. One court found training on books was fair use but keeping pirated copies was not, producing a $1.5 billion settlement worth about $3,000 per work.

By the UISC BD Editorial Desk · United Information Service Center · Published 13 September 2026 · 7-minute read

Artificial intelligence companies face a wave of litigation from authors, artists and media organisations over the unlicensed use of copyrighted work to train their models. Trackers now count around 130 AI copyright lawsuits across US and international courts.

This is not legal advice. It is a map of where the law stands, because both creators and everyday AI users are affected.

The Ruling That Drew the First Line

The most consequential decision so far split the question in two.

A court ruled that training AI on copyrighted books constituted fair use — but that storing pirated copies did not. The outcome was a $1.5 billion settlement, amounting to roughly $3,000 per work.

The distinction matters. The ruling did not say AI companies may use anything freely. It said the act of learning from a work can be lawful while the way the work was obtained can still be unlawful.

Outputs Are a Separate Question

On 12 March 2026, Judge Valerie Chen ruled that AI-generated outputs based on copyrighted training data may constitute infringement if they are "substantially similar" to protected works.

So even where training is permitted, an AI that reproduces a protected work closely enough can still infringe.

The Cases Still Open

The New York Times v. OpenAI. The Times argues ChatGPT can reproduce its articles nearly verbatim. The case was still ongoing as of April 2026, and its appeal sits in the Second Circuit.

Authors v. Meta. A class action alleges Meta trained its Llama models on pirated book datasets. Clearly pirated training material is considered the weakest fair-use position for AI companies.

The appeals courts. The Ninth Circuit is positioned to be the first appellate court to rule directly on training as fair use, with the Second Circuit close behind on the Times appeal and the consolidated Authors Guild action. Those rulings will shape the law far more than any single trial.

Who Owns What AI Makes?

This is the question most everyday users actually have.

In many legal systems, copyright protection depends on human authorship. Output produced purely by a machine, with little human creative input, may not be protectable at all — meaning you may not be able to stop others copying it.

The more substantial your own creative contribution — selecting, arranging, editing, combining — the stronger your claim. Rules differ between countries and are still developing, so treat important commercial work with care.

Practical Guidance

If you use AI to create content:

• Do not ask it to imitate a named living artist or reproduce a specific work.
• Check the provider's terms for commercial use — they differ.
• Add real creative work of your own if ownership matters to you.
• Label AI-generated images, as described in our image generation guide.

If you create original work:

• Keep dated records of your drafts and originals.
• Review the terms of platforms where you publish; some grant AI training rights.
• Follow collective actions in your field. Settlements in the book cases paid individual rights-holders.

Why It Matters Beyond America

Most of these cases are in US courts, but their outcomes set the terms on which global AI companies operate everywhere.

For Bangladesh's creative economy — animation studios, streaming content makers, filmmakers and designers — the question is both defensive and commercial. Protecting original work matters. So does knowing which AI tools can be used safely in paid client projects.

Related reading

• Startups Raised $510 Billion in Six Months, More Than All of Last Year
• Vibe Coding: Why AI-Written Code Fails Security Tests
• Is Your Data Safe With AI Chatbots? How to Opt Out of Training
• Anthropic Has Overtaken OpenAI in Business AI Spending

Sources

• "AI in litigation series: an update on AI copyright cases in 2026," Norton Rose Fulbright — nortonrosefulbright.com
• "AI copyright lawsuits (2026): 130 cases tracked," AI Lawsuit Tracker — ailawsuittracker.com
• "AI copyright lawsuits heat up: March 2026 court rulings and industry takeaways," Tech Daily Shot — techdailyshot.com
• "AI copyright training data 2026: status, timeline, risk," AI Vortex — aivortex.io