BY SRH

Tech Companies Are Accused of Stealing the Work of News Organizations, Authors, and Musicians in Order to Train Their Artificial Intelligence Models…

Content producers have filed dozens of lawsuits against AI developers, alleging they stole their work to profit and create products that may replace their businesses.

News outlets, novelists, photographers, and singers sued, alleging that the firms utilized their content to train and build AI tools without paying them and racing to replace their creative labor.

An online tracker by Santa Clara Law professor Edward Lee shows over 100 such cases under discovery or pre-discovery while litigants prepare evidence for court.
Others face summary judgment, when the judge decides without a trial.

These cases highlight another concern: that the rapidly increasing AI sector and its potential to kill mankind constitute an existential danger to traditional content creation enterprises.

Overall, they highlight the conflict between modern technology’s benefits and the potential loss of human-driven information collection and creative output.
The Doom Loop

News companies and others claim that AI firms undermine their business by leveraging their high-quality, time-, effort-, and money-intensive labor without compensation.

The New York Times and more than a dozen news outlets sued, saying defendants wanted to free-ride on The Timer’s vast investment in journalism by building substitutive goods without permission or payment.

Our AI content approach has created a ‘doom loop’ that will harm our models and the web.
Microsoft director of applied science Brent Hecht, quoted in court docs

Newly leaked litigation docs against OpenAI and Microsoft suggest the firms knew their tactics could be harmful.
“Millions of people around the world will soon consider large models ‘hoovering up’ all their work to be an astonishing theft of unprecedented proportions,” Microsoft Director of Applied Science Brent Hecht remarked.

New York Times Building on Feb. 5, 2024. The media organization and over a dozen others are suing OpenAI and Microsoft for exploiting journalistic information to train AI models without compensation or permission.Samira Bouaou/Epoch Times

In another court-cited internal document, Hecht claimed, “Our AI content strategy has started a ‘doom loop’ that will hurt the performance of our models and the entire web at the same time.”

Starting with The Associated Press in July 2023, OpenAI has sought licenses from news outlets used to train its algorithms. News Corporation, The Washington Post, Politico, and others signed on.

Defendants want to use The Times’s huge news investment to construct substitute products without permission or payment.
New York Times v. Microsoft/OpenAI copyright case

Companies that refused license agreements sued.

Meta made similar deals but was sued by unagreed-upon authors. After individual plaintiffs sued Google for harvesting private user data from across the internet, the firm teamed with Reddit to train its model.

Anthropic has defended itself in court using the fair use theory without licensing deals.

On Sept. 29, 2026, Open AI CEO Sam Altman gives the keynote address at the OpenAI developers conference in San Francisco; on Oct. 7, 2026, Microsoft Copilot executive vise president Jacob Andreou speaks at a Windows event in San Francisco; on April 29, 2025, Meta Chief Product Officer Chris Cox speaks at LlamaCon 2025, an AI developer conference in Menlo Park, Calif. Heather Diehl/Getty Images, Jeff Chiu/AP Photo/File, Benjamin Fanjoy/Getty Images Fair Use?

Companies must feed their AI models a lot of data to teach them to respond to user prompts. That knowledge comes from somewhere, and much of it is copyrighted.

Companies claim that using copyrighted text to train AI models is permissible under Section 107 of the Copyright Act’s fair use concept.

Fair use enables copyright exceptions where necessary. Copyrighted material can be used for research, teaching, or criticism.

AI businesses claim that Section 107 of the Copyright Act’s fair use theory protects pulling copyrighted text to train AI models.

Kevin Frazier, University of Texas–Austin director of AI innovation and law, told Us that the exception depends on how copyrighted content is handled.

He explained the rationale for that mostly depends on the first aspect judges evaluate when determining fair use, which is this nebulous but crucial concept of ‘transformational use.’

Fair use is more likely for AI companies that “scrape” the internet, books, or other sources for text to train their models, Frazier added.

A screen displays stylist prompts and AI-generated shoe models at Imki’s offices in Schiltigheim, France, on Feb. 26, 2025. Custom generative AI solutions for luxury, fashion, and culture are its specialty. “The result is this highly complex, highly nuanced, highly sophisticated AI model that is wildly different from the original work that was used to train the model itself, and so many people regard it as transformative.” Meta defeated a lawsuit by 13 authors who claimed Meta utilized their material to teach “Llama.”

The judge found that using their books to train the model was “highly transformative” and that Meta’s output did not affect their work’s market value.
A judge found Meta’s use of books to train Llama, its AI model, ‘highly transformative’ in 2025 and did not affect authors’ market value.

In early September, the U.S. Department of Justice argued that AI is vital to national security and that the government had a “sovereign interest” in supporting innovation.

The government wants courts to allow AI companies to acquire copyrighted data under fair use. The DOJ said “only the largest technology companies might have the capital necessary to pay licensing fees.” This is needed to encourage smaller enterprises to compete.

AI Copyright Case Tracker

Top staff at OpenAI and Microsoft MSFT -1.93%decrease; down pointing triangle knew that scraping news websites to train their artificial intelligence models may eventually damage or even destroy those publications. They did it anyway, according to a legal filing unsealed Thursday in a high-stakes copyright infringement case.