Copyright Law in 2025: Courts begin to draw lines around AI training, piracy, and market harm; Reuters, March 16, 2026

Edward D. Lanquist and Benjamin West Janke , Reuters; Copyright Law in 2025: Courts begin to draw lines around AI training, piracy, and market harm

"In 2025, U.S. courts issued the first substantive, merits-stage decisions addressing whether the use of copyrighted works to train generative artificial intelligence systems constitutes "fair use." Although these rulings do not settle all open questions — and in some respects highlight emerging judicial disagreements — they represent a significant inflection point in copyright law's response to large language models, image generators, and other foundation models.

Taken together, these cases establish early guideposts for AI developers, publishers, media companies, and enterprises deploying generative AI systems. Below, we summarize the most important copyright ​decisions and pending cases shaping the law in 2025...

Conclusion and recommendations

The ​2025 decisions reflect cautious but meaningful progress in defining how copyright law applies to generative AI. Courts are increasingly receptive to fair use arguments for training on lawfully acquired data, deeply skeptical of speculative market-harm claims, and uniformly intolerant of piracy. At the same time, cases involving direct competition, news content, and human likeness may test the limits of these early rulings."

Axios AI+DC Summit: Copyright protection in the AI era will be up to the courts, industry leaders say; Axios, March 27, 2026

 Julie Bowen, Axios ; Axios AI+DC Summit: Copyright protection in the AI era will be up to the courts, industry leaders say

"Washington, D.C. — As policymakers grapple with how to regulate AI, the hardest questions around copyright and fair use are being punted to the courts, according to governance, creator, and technology experts at an Axios expert voices roundtable.

The big picture: With Congress moving slowly and disagreements over policy, judges are becoming the primary deciders of how AI and the creators work together — or don't.

That's partly by necessity: "Fair use is incredibly complicated — case by case, fact specific," News/Media Alliance president and CEO Danielle Coffey said.

"Each case that we get … we start to get these new guideposts," Jones Walker partner Graham Ryan said.

Ryan said they expect at least three fair use decisions this year that will have implications for the broader AI-artist ecosystem.

Axios' Maria Curi and Ashley Gold moderated the March 25 discussion, which was sponsored by Adobe.

What they're saying: Legal uncertainty remains. For example, two courts within the same district, and during the same week, differed in the reasoning behind their rulings on similar matters of fair use and AI.

"There is a current, live controversy over … the extant understanding of the fourth factor in fair use, which is: Does the copy replace the market for the work?" said Kevin Bankston, senior adviser for the Center for Democracy & Technology.

Still, "we have been trying to support the process through the courts, because we think there is a really strong framework in copyright law for protecting artists right now," according to Public Knowledge president and CEO Chris Lewis."

Can coding agents relicense open source through a “clean room” implementation of code?; Simon Willison's Weblog, March 5, 2026

 Simon Willison's Weblog ; Can coding agents relicense open source through a “clean room” implementation of code?

"Can a model trained on a codebase produce a morally or legally defensible clean-room implementation?"

Everybody’s Talking About AI: Takeaways from the February 20, 2026 Fordham Law Symposium; Lexology, February 26, 2026

Seyfarth Shaw LLP - Owen Wolfe, Lexology; Everybody’s Talking About AI: Takeaways from the February 20, 2026 Fordham Law Symposium

"On February 20, 2026, Gadgets, Gigabytes and Goodwill Blog co-editor Owen Wolfe spoke at the Fordham School of Law as part of the Fordham Intellectual Property, Media & Entertainment Law Journal Symposium, The Meaning of Ownership: Rethinking Intellectual Property, Creativity, and Control in the Age of Innovation. Owen discussed how courts have so far applied the “fair use” doctrine to cases involving generative AI, distinguishing between use of copyrighted materials in gen AI training and gen AI outputs that are alleged to be substantially similar to the original works. He noted that the decisions to date have been mixed, with some courts finding that certain uses of copyrighted works for AI training are fair use, and other courts expressing skepticism about whether that is the correct result. Owen also surveyed arguments both for and against a finding of fair use, giving the audience food for thought about what courts might decide in the future and whether we might see an amendment to the Copyright Act down the road.

Owen’s talk followed one by Dr. Douglas Lind, a professor at Virginia Tech, who surveyed the history of copyright law in the United States. He focused on the law’s treatment of phonograph records and sound recordings when those new technologies first emerged. Dr. Lind noted that copyright law evolved, and the Copyright Act was eventually amended, to address those new technologies. Dr. Lind raised the question of whether the Copyright Act should be amended again to address gen AI."

How ‘Nirvanna the Band the Show the Movie’ pushed the limits of copyright law to create its most WTF moments; Page Six Hollywood, February 14, 2026

 Katcy Stephan, Page Six Hollywood; How ‘Nirvanna the Band the Show the Movie’ pushed the limits of copyright law to create its most WTF moments

"“Nirvanna the Band the Show the Movie,” the latest mockumentary collaboration between director Matt Johnson and composer Jay McCarrol, probably shouldn’t exist. The film features extended parodies that carefully skirt copyright law, a stunt that sees the duo literally dive off Toronto’s CN tower and at least two dozen more moments that will leave audiences asking “How the hell is this legal?” 

It’s a miracle the film made it to screens at all."

MPA Calls On TikTok Owner ByteDance To Curb New AI Model That Created Tom Cruise Vs. Brad Pitt Deepfake; Deadline, February 12, 2026

Ted Johnson , Deadline; MPA Calls On TikTok Owner ByteDance To Curb New AI Model That Created Tom Cruise Vs. Brad Pitt Deepfake

"As reported by Deadline’s Jake Kanter, Seedance 2.0 users are prompting the Chinese AI tool to create videos that appear to be repurposing, with startling accuracy, copyrighted material from studios, including Disney, Warner Bros Discovery and Paramount. In addition to the Cruise vs. Pitt fight, the model has produced remixes of Avengers: Endgame and a Friends scene in which Rachel and Joey are played by otters."

When AI and IP Collide: What Journalists Need to Know; National Press Foundation (NPF), January 22, 2026

National Press Foundation (NPF); When AI and IP Collide: What Journalists Need to Know

"With roughly 70 federal lawsuits waged against AI developers, the intersection of technology and intellectual property is now one of the most influential legal beats. Courts are jumping in to define the future of “fair use.” To bridge the gap between complex legal proceedings and the public’s understanding, NPF held a webinar to unpack these intellectual property battles. One thing all of the expert panelists agreed on: most cases are either an issue of input – i.e. what the AI models pull in to train on – or output – what AI generates, as in the case of Disney and other Hollywood studios v. Midjourney.

“The behavior here of AI companies and the assertion of fair use is completely understandable in our market capitalist system – all players want something very simple. They want their inputs for little or nothing and their outputs to be very expensive,” said Loyola Law professor Justin Hughes. “The fair use argument is all about AI companies wanting their inputs to be free, just like ranchers want their grazing land from the federal government to be free or their mining rights to be free.” AI Copyright Cases Journalists Should Know: Bartz et al. v. Anthropic: Anthropic reached a $1.5 billion settlement in a landmark case for the industry after a class of book authors accused the company of using pirated books to train the Claude AI model. “The mere training itself may be fair use, but the retention of these large copy data sets and their replication or your training from having taken pirated data sets, that’s not fair use,” Hughes explained. The NYT Company v. Microsoft Corporation et al.: This is a massive multi-district litigation in New York where the NYT is suing OpenAI. The Times has pushed for discovery into over 20 million private ChatGPT logs to prove that this model is being used to get past paywalls. Advance Local Media LLC et al. v. Cohere Inc.: The case against the startup Cohere is particularly vital for newsrooms as a judge ruled that AI-generated summaries infringe of news organizations’ ability to get traffic on their sites. “We’ve seen, there’s been a lot of developers who have taken the kind of classic Silicon Valley approach of ask forgiveness rather than permission,” said Terry Hart, general counsel of the Association of American Publishers. “They have gone ahead and trained a lot of models using a lot of copyrighted works without authorization.” Tech companies have trained massive models to ingest the entirety of the internet, including articles, without prior authorization, and Hughes points out that this is a repeated occurrence. AI companies often keep unauthorized copies of these vast datasets to retrain and tweak their models, leading to multiple steps of reproduction that could violate copyright. AI and U.S. Innovation A common defense from tech companies in Silicon Valley is that using these vast amounts of data is necessary to U.S. innovation and keeping the economy competitive. “‘We need to beat China, take our word for it, this is going to be great, and we’re just going to cut out a complete sector of the economy that’s critical to the success of our models,'” Hart said. “In the long run, that’s not good for innovation. It’s not good for the creative sectors and it’s not good for the AI sector.” Reuters technology reporter Deepa Seetharaman has also heard the China competition argument, among others. “The metaphor that I’ll hear a lot here is, ‘it’s like somebody visiting a library and reading every book, except this is a system that can remember every book and remember all the pieces of every book. And so why are you … harming us for developing something that’s so capable?'” Seetharaman said. Hughes noted that humans are not walking into a library with a miniature high-speed photocopier to memorize every book. Humans don’t memorize with the “faithful” precision of a machine. Hart added that the metaphor breaks down because technology has created a new market space that isn’t comparable to a human reader. Speakers:

• Wayne Brough, Resident Senior Fellow, Technology and Innovation Team, R Street
• Terry Hart, General Counsel, Association of American Publishers
• Justin Hughes, Honorable William Matthew Byrne Distinguished Professor of Law, Loyola Marymount University
• Deepa Seetharaman, Tech Correspondent, Reuters

Summary and transcript: https://nationalpress.org/topic/when-... This event is sponsored by The Copyright Alliance and NSG Next Solutions Group. This video was produced within the Evelyn Y. Davis studios. NPF is solely responsible for the content."

Search Engines, AI, And The Long Fight Over Fair Use; Electronic Frontier Foundation (EFF), January 23, 2026

JOE MULLIN , Electronic Frontier Foundation (EFF); Search Engines, AI, And The Long Fight Over Fair Use

"We're taking part in Copyright Week, a series of actions and discussions supporting key principles that should guide copyright policy. Every day this week, various groups are taking on different elements of copyright law and policy, and addressing what's at stake, and what we need to do to make sure that copyright promotes creativity and innovation.

Long before generative AI, copyright holders warned that new technologies for reading and analyzing information would destroy creativity. Internet search engines, they argued, were infringement machines—tools that copied copyrighted works at scale without permission. As they had with earlier information technologies like the photocopier and the VCR, copyright owners sued.

Courts disagreed. They recognized that copying works in order to understand, index, and locate information is a classic fair use—and a necessary condition for a free and open internet.

Today, the same argument is being recycled against AI. It’s whether copyright owners should be allowed to control how others analyze, reuse, and build on existing works."

How researchers got AI to quote copyrighted books word for word; Le Monde, January 24, 2026

Nicolas Six , Le Monde; How researchers got AI to quote copyrighted books word for word

"Where does artificial intelligence acquire its knowledge? From an enormous trove of texts used for training. These typically include vast numbers of articles from Wikipedia, but also a wide range of other writings, such as the massive Books3 dataset, which aggregates nearly 200,000 books without the authors' permission. Some proponents of conversational AI present these training datasets as a form of "universal knowledge" that transcends copyright law, adding that, protected or not, AIs do not memorize these works verbatim and only store fragmented information.

This argument has been challenged by a series of studies, the latest of which, published in early January by researchers at Stanford University and Yale University, is particularly revealing. Ahmed Ahmed and his coauthors managed to prompt four mainstream AI programs, disconnected from the internet to ensure no new information was retrieved, to recite entire pages from books."

FREE WEBINAR: REGISTER: AI, Intellectual Property and the Emerging Legal Landscape; National Press Foundation, Thursday, January 22, 2026

National Press Foundation; REGISTER: AI, Intellectual Property and the Emerging Legal Landscape

"Artificial intelligence is colliding with U.S. copyright law in ways that could reshape journalism, publishing, software, and the creative economy.

The intersection of AI and intellectual property has become one of the most consequential legal battles of the digital age, with roughly 70 federal lawsuits filed against AI companies and copyright claims on works ranging from literary and visual work to music and sound recording to computer programs. Billions of dollars are at stake.

Courts are now deciding what constitutes “fair use,” whether and how AI companies may use copyrighted material to build models, what licensing is required, and who bears responsibility when AI outputs resemble protected works. The legal decisions will shape how news, art, and knowledge are produced — and who gets paid for them.

To help journalists better understand and report on the developing legal issues of AI and IP, join the National Press Foundation and a panel of experts for a wide-ranging discussion around the stakes, impact and potential solutions. Experts in technology and innovation as well as law and economics join journalists in this free online briefing 12-1 p.m. ET on Thursday, January 22, 2026."

Copyright laws need to modernize to include fan-made edits; The Miami Hurricane, January 14, 2026

Marissa Levinson, The Miami Hurricane; Copyright laws need to modernize to include fan-made edits

[Kip Currier: Unfortunately, this University of Miami's student newspaper's Op-Ed -- Copyright laws need to modernize to include fan-made edits -- contains staggeringly inaccurate interpretations and assertions about copyright law and fair use. 

No one wanting to make informed decisions on copyright-related matters should rely on the writer's cherry-picked aspects of copyright law that are then stitched together to make wildly erroneous conclusions.

Copyright literacy is essential.]

 

[Excerpt]

"I can’t even count the amount of times I’ve been scrolling through my saved folders on TikTok, Instagram or X to watch video edits of clips from my favorite TV shows, only to find nothing but a shell of what once used to be there, with a body of text over it. It reads: “This video is no longer available due to a copyright claim.” 

Video edits of shows and movies, which are often paired with trending songs as the audios, have gained traction on social media platforms among hundreds of fan bases. Navigating this new age of fan-generated edits comes with confusion. As copyright laws based on precedent aren’t current enough to guide regulations on this new type of content, video edits deserve to be protected under copyright law.

Are edits legal?

Fan-made video edits range from less than 30-seconds to a few minutes long. This poses the question of whether they are legal in terms of copyright. The law gives copyright the “power ‘to promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Investors the exclusive Right to their Respective Writings and Discoveries,’” according to Article 1, Section 8 of the United States Constitution. 

This means, under the Copyright Act of 1976, “original works of authorship fixed in tangible medium of expression” are protected. According to the four exemptions of the copyright law, video edits are protected by the first amendment and therefore, should be legal to publish."

AI’S MEMORIZATION CRISIS: Large language models don’t “learn”—they copy. And that could change everything for the tech industry.; The Atlantic, January 9, 2026

 Alex Reisner, The Atlantic; AI’S MEMORIZATION CRISIS: Large language models don’t “learn”—they copy. And that could change everything for the tech industry

"On tuesday, researchers at Stanford and Yale revealed something that AI companies would prefer to keep hidden. Four popular large language models—OpenAI’s GPT, Anthropic’s Claude, Google’s Gemini, and xAI’s Grok—have stored large portions of some of the books they’ve been trained on, and can reproduce long excerpts from those books."

Extracting books from production language models; Cornell University, January 6, 2026

Ahmed Ahmed, A. Feder Cooper, Sanmi Koyejo, Percy Liang, Cornell University; Extracting books from production language models

"Many unresolved legal questions over LLMs and copyright center on memorization: whether specific training data have been encoded in the model's weights during training, and whether those memorized data can be extracted in the model's outputs. While many believe that LLMs do not memorize much of their training data, recent work shows that substantial amounts of copyrighted text can be extracted from open-weight models. However, it remains an open question if similar extraction is feasible for production LLMs, given the safety measures these systems implement. We investigate this question using a two-phase procedure: (1) an initial probe to test for extraction feasibility, which sometimes uses a Best-of-N (BoN) jailbreak, followed by (2) iterative continuation prompts to attempt to extract the book. We evaluate our procedure on four production LLMs -- Claude 3.7 Sonnet, GPT-4.1, Gemini 2.5 Pro, and Grok 3 -- and we measure extraction success with a score computed from a block-based approximation of longest common substring (nv-recall). With different per-LLM experimental configurations, we were able to extract varying amounts of text. For the Phase 1 probe, it was unnecessary to jailbreak Gemini 2.5 Pro and Grok 3 to extract text (e.g, nv-recall of 76.8% and 70.3%, respectively, for Harry Potter and the Sorcerer's Stone), while it was necessary for Claude 3.7 Sonnet and GPT-4.1. In some cases, jailbroken Claude 3.7 Sonnet outputs entire books near-verbatim (e.g., nv-recall=95.8%). GPT-4.1 requires significantly more BoN attempts (e.g., 20X), and eventually refuses to continue (e.g., nv-recall=4.0%). Taken together, our work highlights that, even with model- and system-level safeguards, extraction of (in-copyright) training data remains a risk for production LLMs."

AI copyright battles enter pivotal year as US courts weigh fair use; Reuters, January 5, 2026

 Blake Brittain , Reuters; AI copyright battles enter pivotal year as US courts weigh fair use

"The sprawling legal fight over tech companies' vast copying of copyrighted material to train their artificial intelligence systems could be entering a decisive phase in 2026.

After a string of fresh lawsuits and a landmark settlement in 2025, the new year promises to bring a wave of rulings that could define how U.S. copyright law applies to generative AI. At stake is whether companies like OpenAI, Google and Meta can rely on the legal doctrine of fair use to shield themselves from liability – or if they must reimburse copyright holders, which could cost billions."

‘Pirated’?: NY Defenders Face Copyright Suit for Allegedly ‘Copying’ Expert Report; Law.com, January 2, 2026

Alyssa Aquino , Law.com; ‘Pirated’?: NY Defenders Face Copyright Suit for Allegedly ‘Copying’ Expert Report

"A researcher in their complaint alleged that the Federal Defenders of New York copied an expert report commissioned by other attorneys directly into court filings for their own case. “When you’re using something in litigation, you usually have a fair use defense, but that’s usually because you’re using it for something different than its original purpose,” said Stacey Lantagne, a law professor at Suffolk University. “But here, [the report] seems to have been created solely for litigation.”"

Artificial Intelligence, Copyright, and the Fight for User Rights: 2025 in Review; Electronic Frontier Foundation (EFF), December 25, 2025

 TORI NOBLE, Electronic Frontier Foundation (EFF); Artificial Intelligence, Copyright, and the Fight for User Rights: 2025 in Review

"A tidal wave of copyright lawsuits against AI developers threatens beneficial uses of AI, like creative expression, legal research, and scientific advancement. How courts decide these cases will profoundly shape the future of this technology, including its capabilities, its costs, and whether its evolution will be shaped by the democratizing forces of the open market or the whims of an oligopoly. As these cases finished their trials and moved to appeals courts in 2025, EFF intervened to defend fair use, promote competition, and protect everyone’s rights to build and benefit from this technology.

At the same time, rightsholders stepped up their efforts to control fair uses through everything from state AI laws to technical standards that influence how the web functions. In 2025, EFF fought policies that threaten the open web in the California State Legislature, the Internet Engineering Task Force, and beyond."

Fair Use is a Right. Ignoring It Has Consequences.; Electronic Frontier Foundation (EFF), December 18, 2025

MITCH STOLTZ , Electronic Frontier Foundation (EFF); Fair Use is a Right. Ignoring It Has Consequences.

"Fair use is not just an excuse to copy—it’s a pillar of online speech protection, and disregarding it in order to lash out at a critic should have serious consequences. That’s what we told a federal court in Channel 781 News v. Waltham Community Access Corporation, our case fighting copyright abuse on behalf of citizen journalists."

(Podcast) The Briefing: What Is Fair Use and Why Does It Matter? (Featured); JDSupra, December 5, 2025

Richard Buckley, Jr. and Scott Hervey, JDSupra ; (Podcast) The Briefing: What Is Fair Use and Why Does It Matter? (Featured)

"Creators, beware: just because it’s online doesn’t mean it’s fair game. In this episode of The Briefing, Scott Hervey and Richard Buckley break down one of the most misunderstood areas of copyright law—fair use.

In this featured episode, they cover:

- What makes a use “transformative”?

- Why credit alone doesn’t protect you

- How recent court rulings (Warhol v. Goldsmith) are changing the game

- Tips to stay on the right side of the law"

The Disney-OpenAI Deal Redefines the AI Copyright War; Wired, December 11, 2025

BRIAN BARRETT, Wired; The Disney-OpenAI Deal Redefines the AI Copyright War

 "“I think that AI companies and copyright holders are beginning to understand and become reconciled to the fact that neither side is going to score an absolute victory,” says Matthew Sag, a professor of law and artificial intelligence at Emory University. While many of these cases are still working their way through the courts, so far it seems like model inputs—the training data that these models learn from—are covered by fair use. But this deal is about outputs—what the model returns based on your prompt—where IP owners like Disney have a much stronger case

Coming to an output agreement resolves a host of messy, potentially unsolvable issues. Even if a company tells an AI model not to produce, say, Elsa at a Wendy’s drive-through, the model might know enough about Elsa to do so anyway—or a user might be able to prompt their way into making Elsa without asking for the character by name. It’s a tension that legal scholars call the “Snoopy problem,” but in this case you might as well call it the Disney problem.

“Faced with this increasingly clear reality, it makes sense for consumer-facing AI companies and entertainment giants like Disney to think about licensing arrangements,” says Sag."

Film Studios, News Media and Even Competitor LexisNexis Among the Nine Amicus Briefs Supporting Thomson Reuters’ Copyright Case Against ROSS; LawSites, December 8, 2025

Bob Ambrogi , LawSites ; Film Studios, News Media and Even Competitor LexisNexis Among the Nine Amicus Briefs Supporting Thomson Reuters’ Copyright Case Against ROSS

"The long-running copyright litigation between Thomson Reuters and ROSS Intelligence is now pending in the 3rd U.S. Circuit Court of Appeals for an interlocutory appeal of the trial judge’s rulings in favor of TR. 

Recently here, I reported on the 10 amicus curiae briefs filed in support of ROSS, all arguing that the now-defunct AI legal research startup did not violate copyright law. 

Now, nine amicus briefs have been filed in support of TR. Those filing briefs range from major movie studios such as Disney and Paramount, to news media and copyright organizations, to individual copyright law professors, and even to TR’s principal competitor LexisNexis."