Generative AI and journalism: who is suing and who is cutting deals?

Generative artificial intelligence has turned the global media market into a full-blown battleground. On one side, news organizations are taking tech companies to court over the unauthorized use of their content. On the other, publishers are closing multimillion-dollar deals with those very same companies to license what they produce. This tension between content creators and those training AI models is not exactly new, but it has reached a staggering scale in recent months, moving billions of dollars and redefining how the journalism industry positions itself in the face of technology.

Since late 2023, OpenAI has racked up more than a dozen lawsuits filed by media organizations around the world — and Perplexity AI is not far behind, with roughly five ongoing cases. At the same time, deals keep being announced at a rapid pace, involving major names like News Corp, the Financial Times, the Guardian, and hundreds of other publications. These are two opposite paths in response to the same problem: the value of journalism in the age of generative AI still does not have a defined price — and everyone is racing to get there first. 🤔

What is at stake goes far beyond settlements or contracts. This clash will shape how digital journalism operates in the years ahead — and who gets to set the rules of the game.

The lawsuits that put OpenAI on the stand

The most iconic case in this legal war was filed by The New York Times in December 2023, when the newspaper sued OpenAI and Microsoft, alleging that their artificial intelligence models were trained on millions of articles published by the outlet without any authorization or financial compensation. The lawsuit sought damages and the destruction of all language models trained on the newspaper’s content. The NYT and OpenAI had negotiated for nine months before the case was filed, but the publication concluded that no resolution was moving forward and decided to take the matter to court.

In response, OpenAI argued that nobody uses ChatGPT as a substitute for a NYT subscription and that the newspaper had paid someone to hack its products, running tens of thousands of attempts to generate results where entire paragraphs from articles were reproduced verbatim. According to OpenAI, this was only possible by exploiting a specific bug and using prompts that violated the platform’s terms of service.

Soon after, other publishers followed the same path. Eight daily newspapers owned by Alden Global Capital — including the New York Daily News, Chicago Tribune, Denver Post, and Orlando Sentinel — sued OpenAI and Microsoft in April 2024. In November 2025, nine more regional newspapers from the same holding company filed a new lawsuit, collectively seeking more than 10 billion dollars in damages.

The Chicago Tribune also sued Perplexity separately in late 2025, accusing the company of illegally copying millions of stories, videos, images, and other copyrighted content to feed its answer engine. The lawsuit pointed out that Perplexity generates content identical or substantially similar to the originals, and that when it does not copy, it produces hallucinations — fabricated information falsely attributed to the newspaper.

International lawsuits gain momentum

The movement did not stay confined to the United States. In Denmark, the organization DPCMO filed a lawsuit against OpenAI in February 2026, alleging that the company trained its models on content from Danish publishers without giving them the option to opt out. According to the group, OpenAI refused negotiations and did not participate in mediations proposed by the Danish government.

In Brazil, Folha de S.Paulo sued OpenAI in August 2025, arguing that the company accesses the newspaper’s website daily by circumventing protective mechanisms and distributes its content to users, siphoning traffic away from the outlet. The newspaper demanded that OpenAI stop using its content without authorization and destroy all models trained on copyrighted material.

In Japan, Yomiuri Shimbun, one of the country’s largest newspapers, sued Perplexity in August 2025 over the unauthorized acquisition of more than 119,000 articles between February and June of that year. The outlet sought damages of approximately 9.9 million pounds and compensation for lost advertising revenue, since Perplexity users had begun consuming summaries instead of clicking through to the original links.

In India, outlets such as Indian Express, Hindustan Times, and NDTV requested to join an existing lawsuit against OpenAI. And in Canada, a coalition of major publishers including CBC/Radio-Canada, Toronto Star, Globe and Mail, and Postmedia filed a class-action lawsuit in November 2024, asserting that OpenAI routinely violates copyright and terms of use by scraping large volumes of content to develop its products. 😬

Perplexity under growing pressure

Perplexity AI has become a frequent target of lawsuits and legal threats. Beyond the cases filed by the NYT, Chicago Tribune, News Corp, and Yomiuri Shimbun, Encyclopedia Britannica and Merriam-Webster also sued the company in September 2025, alleging that Perplexity free-rides on their content by summarizing articles and redirecting traffic that would otherwise go to their sites. The same entities filed another lawsuit against OpenAI in March 2026, with similar allegations of mass copying and verbatim reproduction of content.

Even the BBC, though without a formal lawsuit, threatened to sue Perplexity after presenting evidence that the company’s default model was trained on BBC content and that search results included verbatim material and recent links from the BBC website. In response, Perplexity called the claims manipulative and opportunistic.

The Penske case against Google

One especially noteworthy lawsuit was filed by Penske Media Corp — parent company of the Hollywood Reporter, Variety, Deadline, and Rolling Stone — against Google. Unlike the other cases, this one does not involve model training: Penske sued Google over the impact of AI Overviews on search results, alleging that roughly 20% of searches linking to its sites display AI-generated summaries, drastically reducing clicks. The company reported a decline of more than a third in affiliate link revenue compared to late 2024. It is the first lawsuit from a publisher against Google specifically over this AI search feature.

Getty Images and the frustrated precedent

The Getty Images case against Stability AI in the United Kingdom also deserves attention, although with a different outcome than expected. Getty alleged that Stability AI illegally copied and processed millions of its images to train the Stable Diffusion model. However, the case ended up being called disappointing after Getty withdrew the training-related claim during trial and the judge ruled largely in favor of Stability AI. Getty continues its case in the United States.

The other side of the coin: deals worth billions

While part of the journalism industry is betting on the courts, another part decided to sit at the table and negotiate. OpenAI has been especially active in building partnerships with major media organizations, and the numbers are significant.

The deal with News Corp, which controls the Wall Street Journal, New York Post, The Times, and The Sun, was valued at more than 250 million dollars over five years. More recently, News Corp signed another contract with Meta worth up to 50 million dollars per year for at least three years, allowing Meta to use content from News Corp titles in the United States and the United Kingdom for Facebook’s AI tools.

The Financial Times was the first major British outlet to announce a deal with OpenAI, in April 2024. The partnership involves current content and archival material, and the FT also signed agreements with Prorata.ai and joined Google’s AI pilot program.

The Guardian closed a deal with OpenAI in February 2025, securing compensation for the use of its journalism in ChatGPT through short summaries and article excerpts with proper credit. The outlet also signed partnerships with Prorata.ai and is one of the publishers involved in Google’s pilot program.

The explosion of deals across multiple platforms

The list of publishers that have signed deals is extensive and keeps growing. Here are some of the most notable:

• Condé Nast (Vogue, Wired, GQ, Vanity Fair) signed with OpenAI in August 2024 and also with Amazon for the Rufus shopping assistant
• Time signed with OpenAI, granting access to its 101-year archive, and also participates in Perplexity’s revenue-sharing program
• Vox Media (The Verge, New York Magazine, Eater) entered a strategic partnership with OpenAI
• The Atlantic closed deals with both OpenAI and Prorata.ai
• Axel Springer (Politico, Business Insider, Bild) has agreements with both OpenAI and Microsoft
• Associated Press licensed its archive dating back to 1985 to OpenAI and also has a deal with Google to provide real-time information to Gemini
• Hearst (Houston Chronicle, Esquire, Cosmopolitan) signed with OpenAI covering more than 20 magazines and 40 newspapers in the U.S.
• Reuters closed a deal with Meta for real-time use in its AI chatbot and is a Microsoft partner for Copilot
• Reach (Mirror, Express) signed a usage-based contract with Amazon for the Nova AI model and Alexa

Even The New York Times, which leads the lawsuits against OpenAI, closed its first AI licensing deal — but with Amazon, in May 2025, for between 20 and 25 million dollars per year. The agreement allows Amazon to use summaries and excerpts from NYT content, NYT Cooking, and The Athletic in products like Alexa.

Prorata.ai and the revenue-sharing model

One of the most interesting approaches comes from Prorata.ai, which developed a proprietary algorithm to measure how much of each publisher’s content is used in AI-generated responses and allocate compensation proportionally. More than 500 publications have already signed partnerships with the company, including The Atlantic, Fortune, Time, The Guardian, Daily Mail, and Sky News. Prorata promises to split 50% of revenues with content creators. 📊

Google enters the game with a different approach

Google announced its first set of AI-focused deals with publishers in December 2025, but with a distinct approach: these are not licensing agreements. Instead, they are described as extensions of existing commercial partnerships, involving cash payments and expanded display rights. Participating publishers include The Guardian, The Washington Post, the Financial Times, Der Spiegel, El País, and Folha. The pilot program tests features such as AI-generated article summaries on publisher pages in Google News and audio briefings.

What this dispute reveals about the future of digital content

The tension between lawsuits and deals is not just a legal or financial matter — it reflects a structural transformation in how the value of information is perceived and distributed on the internet. For decades, the dominant model was digital advertising: produce content, attract an audience, sell ad space. The arrival of generative artificial intelligence broke that cycle by creating an intermediary layer between the content and the end user — a layer that processes, summarizes, and delivers information without necessarily sending the reader to the original source.

Tools we use daily

Translation

Text Inspection & Clipping

Social Media

Search & Research

Recruitment

Programming

Productivity & Organization

Operations

Marketing & Sales

Finance & Investment

Education

Design & Media

Data & Analytics

Content & Writing

Automation

For OpenAI and similar companies, deals with publishers work as a kind of legal and reputational insurance. Negotiating licenses reduces the risk of lawsuits, demonstrates good faith to regulators, and builds a high-quality dataset for ongoing model training. Verified content produced by professional journalists carries a different weight in training compared to random text scraped from the open web. That explains the real willingness of these companies to sign contracts with significant dollar amounts.

Part of the journalism community, however, sharply criticizes the outlets that chose to negotiate. The argument is that by licensing content to AI systems capable of answering questions directly and comprehensively, publishers would be training their own replacements — systems that inform the public without the user ever needing to visit the outlet’s website, reducing traffic and advertising revenue.

The amounts being offered also spark debate. Reports indicated that OpenAI initially offered between 1 and 5 million dollars per year for publishers to license their content — figures considered low by many outlets. The News Corp deal, valued at more than 250 million dollars over five years, landed well above that range, suggesting that bargaining power varies enormously depending on the size and relevance of the publisher.

Who is suing whom: the full map

The list of lawsuits involving publishers and AI companies is extensive. Among the most notable cases are:

• Encyclopedia Britannica and Merriam-Webster vs. OpenAI and Perplexity
• DPCMO (Danish organization) vs. OpenAI
• The New York Times vs. OpenAI, Microsoft, and Perplexity
• Chicago Tribune vs. Perplexity
• US News & World Report vs. OpenAI
• Alden Global Capital newspapers vs. OpenAI and Microsoft (two separate lawsuits)
• Penske Media Corp vs. Google (AI Overviews)
• Folha de S.Paulo vs. OpenAI
• Yomiuri Shimbun vs. Perplexity
• Reddit vs. Anthropic
• Ziff Davis (CNET, PCMag, IGN) vs. OpenAI
• News/Media Alliance members vs. Cohere
• Indian publishers vs. OpenAI
• Canadian coalition vs. OpenAI
• News Corp vs. Perplexity
• Mumsnet vs. OpenAI
• Center for Investigative Reporting vs. OpenAI and Microsoft
• The Intercept, Raw Story, and AlterNet vs. OpenAI
• Getty Images vs. Stability AI

The landscape taking shape

Defining a fair price for journalistic content in the age of generative AI is, right now, one of the most complex and urgent questions in the digital ecosystem. Ongoing lawsuits may set important legal precedents, but courts tend to move far too slowly to keep pace with the rapid speed of technological development. Deals closed voluntarily create market benchmarks, but they vary enormously in terms of value and conditions, making any standardization difficult.

There are also initiatives trying to build bridges between the two worlds. OpenAI and Microsoft earmarked 10 million dollars for the Lenfest Institute for Journalism to fund AI projects in local newsrooms across the United States. OpenAI also invested 5 million dollars in the American Journalism Project and funded the launch of four new local Axios newsrooms. These moves suggest that AI companies recognize the need to maintain a healthy journalism ecosystem — whether out of conviction, strategy, or regulatory pressure.

What seems certain is that the next two to three years will be decisive in setting the rules of this game. Publishers that understand this dynamic clearly — evaluating when it makes sense to negotiate and when it makes sense to go to court — will have a real advantage in this negotiation. Artificial intelligence is not going away, and quality journalism will remain essential for these models to work well. The central question remains: who will capture the value generated by this relationship? 🚀