Skip to main content
OpenAI

Reporting the facts about the New York Times’ lawsuit

Fair use benefits everyone. The New York Times has tried to present this lawsuit as a binary choice between human authorship and AI. That’s simply wrong. It’s also inconsistent with even their own use of this technology, not to mention the experience of many others within the news and publishing industries who are using AI in ways that keep human authorship front and center. At the heart of this case is the principle of fair use and the role it plays in delivering benefits for everyone; from accelerating medical research, to helping people grow small businesses and giving local newsrooms the tools to reach new audiences. This case is not about human authorship versus AI. It’s about the New York Times looking for an undeserved payday at the expense of progress that benefits everyone.

A win-win: Fair use benefits everyone

We’ve consistently supported journalism, the long-established principles of fair use, and the Constitution’s promise of a more open, competitive future for sharing knowledge. Over the past three years—long before this lawsuit was filed—we’ve worked together with leaders in the news industry to find new ways to benefit from AI. From reaching new audiences, to building interactive tools that make content more accessible, and enhancing business processes, newsrooms around the country are using our technology to support the important work they do to serve their communities. Fair use is central to unlocking these opportunities.

Fair use isn’t some technical legal term that only affects what happens inside a courtroom; it’s played a critical role in decades of American innovation and creativity that’s benefited countless people over the years. Here, it’s fundamentally about facts and knowledge belonging to everyone, for the benefit of everyone.

Today, training AI models on publicly available information is just one of the many, many ways the fair use doctrine boosts innovation. Our models are trained on publicly available data and grounded in fair use. Over a billion people each week use these models to improve their daily lives, such as by navigating the health system, growing their businesses⁠(opens in a new window), learning new skills or even just getting help with everyday life tasks⁠(opens in a new window) so they can spend more time with their families. Many people also use our models in ways that directly benefit the greater public, such as accelerating scientific discovery, improving health care and advancing medical⁠ research. The OpenAI Foundation⁠(opens in a new window) is also focused on ensuring people benefit from AI, including through its work to accelerate life sciences⁠(opens in a new window).

These are just a few examples of how fair use enables innovation that makes a positive and meaningful difference in real people’s lives. The Times is seeking to stand in the way of all of that, all for its own benefit. The Times’ lawsuit asks the Court to ignore existing law and a principle that has long delivered tangible benefits to society, and to allow their desires to take precedence over the interests of the wider public.

As this lawsuit has progressed, the Court has continued to narrow this case⁠(opens in a new window) and dismiss claims, and the Times has been forced to look for new ways to advance its case⁠. You can read more about recent relevant case law here⁠.

Journalism, human authorship and AI

We believe the future of creativity is one that is fundamentally human. For example, a human author might use AI to research a period-accurate castle design for their historical fiction novel, or keep track of character movements across the sprawling fantasy epic they’ve been writing over many years. It’s clear that people want tools that support human creativity, not replace them. To us, success means empowering human authorship and journalism so there’s more of it in the world.

We also believe that journalism plays a vital role in society—a principle that’s very much consistent with our focus on AI that’s built on democratic values. The Times’ own use of our technology in service of a Pulitzer-winning story⁠(opens in a new window) is one of many great examples of the way newsrooms are using these tools to benefit their readers, journalists and businesses.

The New York Times’ business is healthier than ever⁠(opens in a new window). But we are sensitive to the financial challenges that have faced others in the news industry over the past two decades, even though those concerns predated ChatGPT (this report⁠(opens in a new window) provides ⁠(opens in a new window)an example). That’s why we work so hard to further great journalism by providing tools that empower newsrooms and help them build more sustainable businesses. We’ve also led the AI industry by pioneering ways for organizations to express their preferences regarding how OpenAI’s crawlers access their content to appear in search results or train foundation models. AI systems are a force for good in journalism and everyday Americans. The New York Times knows this.

Since 2023, we have partnered with news organizations around the world—both large and small, national and local—to support journalism. From our work with the Lenfest Institute, American Journalism Project and WAN-IFRA, to our partnerships with individual newsrooms to grow local coverage, expand audiences and support the very human work of reporting.

In September 2026, for example, we announced⁠ a new initiative in partnership with WAN-IFRA and the Association of Independent Regional Press Publishers of Ukraine (AIRPPU) to support and strengthen Ukrainian news publishers. As Oksana Brovko, Chief Executive Officer of AIRPPU, said: 

“Ongoing war, economic pressures, and disruption to daily lives have created unprecedented challenges for independent media in Ukraine. We believe that AI can greatly assist and provide significant opportunities to strengthen editorial workflows, improve efficiency, and better serve audiences. This program will provide Ukrainian publishers with access to international expertise, practical tools, and hands-on support that will help them turn those opportunities into tangible results. And most importantly—it should give journalists back time to do journalism.”

You can read more about our work with news organizations here, as well as through our OpenAI Academy for News Organizations hub⁠(opens in a new window).

2026

We filed our motions for summary judgment

The cases continue narrowing

The Court once again dismissed claims in this case, as well as the Ziff Davis case, as both continue proving not to hold up under scrutiny.

2025

The Times invades the privacy of tens of millions of people

As courts have ruled on fair use and their claims in this case have been narrowed, the New York Times increasingly relies on aggressive and unreasonable litigation tactics, such as invading our users’ privacy.

We have been actively fighting the Times’ demand that we turn over 20 million of your private ChatGPT conversations. They claim they might find examples of our users using ChatGPT to try to get around their paywall.

Additional update:

The District Court judge has issued a ruling on our appeal and we have complied with the order, as we are obligated to do. Below, we outline steps we’ve taken to de-identify data and tightly control access to it, as well as our continued efforts to protect user privacy throughout this legal process.

Update on December 16, 2025:

Unfortunately, we are obligated to comply with the Magistrate Judge’s order while we continue to fight this invasion of user privacy by The New York Times in our appeal to the District Court judge.

The data we are making accessible to comply with this order has undergone a de-identification process intended to remove or mask PII and other private information, and is being provided under tight access controls designed to prevent the Times from copying and printing data that isn’t directly relevant to this case. We will continue to object to any attempt by the Times to use chat conversations as part of this litigation in ways that could further harm the privacy of any user.

The Times’ refusal to withdraw their request to violate the privacy of tens of millions of people continues to run counter to the important role journalism has historically played in defending people’s right to privacy. Both the Magistrate Judge and the Times relied on the fact that Anthropic was willing to produce⁠(opens in a new window) five million of their users’ conversations in litigation irrespective of relevance. However, Anthropic’s decision does not establish a precedent here, and does not justify invading the privacy of our users.

Our users’ privacy is a priority and we’ll continue fighting to protect it.

The Times had initially demanded 1.4 billion private ChatGPT conversations be turned over in May 2025. We have continuously pushed back.

Previously, the Times asked the Court to force us to retain all user conversations indefinitely—which even included chats that users chose to delete⁠⁠(opens in a new window). We fought that and restored our users’ control over their private chats. Again, these are users who have nothing to do with the lawsuit. 

AI training is fair use

Two federal judges in two separate cases have already independently confirmed what copyright law has long supported, finding in those cases that training AI models is highly transformative and protected by fair use.

The Conference Board noted⁠(opens in a new window): “These cases are significant victories for developers... In both cases, the judges noted that each AI model’s use of the copyrighted material was highly transformative, a key element of satisfying the fair use doctrine.”

As we’ve been saying:

“AI models are profoundly transformative. They use massive computational power to learn deep mathematical patterns, analyses, and insights from trillions of datapoints so they can create new content—or even ‘think’ in an inner monologue like a person. They are designed to create profound new insights and understandings, and have safeguards to avoid replicating the material they learn from.” —Jason Kwon, Chief Strategy Officer, OpenAI⁠(opens in a new window)

The Times largely avoided informing its readers at the time about these important court decisions, barely covering them in passing. However, others have not shied away from fully reporting on these important rulings, clearly in the public interest—including the Wall Street Journal⁠(opens in a new window), NPR⁠(opens in a new window), Fortune⁠(opens in a new window), The Guardian⁠(opens in a new window), TechCrunch⁠(opens in a new window), and others.

As the case progressed, the Court rejected several of the Times’ claims

The Court continued to scrutinize the Times’ claims and dismissed a number of them, allowing the case to primarily focus on fair use.

The Court also rejected claims brought by Ziff Davis

Similarly, the Court narrowed the case and dismissed a number of claims brought by Ziff Davis, including those related to trademark and DMCA allegations.

2024

During discovery phase, the Times made false claims about data destruction

The Times falsely claimed we “destroyed” data during the discovery process. We explained to the Court what really happened was the Times asked for a setup change on one of the machines we’d given them to review data as part of the legal process. That change, which they requested, wiped the folder structure on a temporary cache drive they were incorrectly using to store their searches. No actual data was lost. All the Times needed to do was re-run their searches.

It also turned out that the Times, which falsely accused us of destroying data, had themselves secretly deleted evidence⁠(opens in a new window) of their extensive use of OpenAI’s models internally. This was a fact that was undisputed in court.

The Court dismisses the related Raw Story Media & Alternet Media case

In a related copyright lawsuit, the judge dismissed Raw Story and AlterNet’s DMCA claims against OpenAI because they didn’t show real, specific harm from OpenAI allegedly removing their copyright information, finding their claims to be too speculative.

The Times introduces references to their legal claims in stories about OpenAI

The Times turned its news coverage into a platform for its legal arguments, embedding lawsuit claims inside stories about OpenAI. While disclosures are common, giving them this level of prominence—often by inserting in the middle of stories—is not.

  • November 2025: (The New York Times has sued⁠(opens in a new window) OpenAI and Microsoft, claiming copyright infringement of news content related to A.I. systems. The companies have denied those claims.)

  • August 2024: (The Times sued⁠(opens in a new window) OpenAI and Microsoft in December for copyright infringement of news content related to A.I. systems.)

Here is an example from a December 2025 piece:

OpenAI, the maker of ChatGPT, said it had extensive safeguards to protect its users’ private information.

A representative noted that users could opt out of having their chats used to train future models, and said the company tested its systems against simulated attacks. It also shares minimal data with third-party service providers, she said. (The Times has sued OpenAI, claiming copyright infringement of news content. OpenAI has denied the claims.)

We file our first legal response to the New York Times’ lawsuit

In a February 2024 motion to dismiss⁠(opens in a new window), we explained how longstanding principles of fair use unequivocally applied to training AI models and is central to this case. We also highlighted the fact that the Times misleadingly went to elaborate lengths to intentionally manipulate prompts in order to manufacture forced regurgitation—a clear misuse of our products that is not how people use ChatGPT.

Meanwhile, across the news industry, leading national and international publications embraced AI to grow and modernize their businesses. Over 20 publications including Axios, The Atlantic, Condé Nast, Dotdash Meredith⁠(opens in a new window), Hearst, News Corp⁠(opens in a new window), Prisa Media, and Vox Media, have partnered with us to use AI to reach new audiences, build new products, and support sustainable journalism.

2023

We set out to support a healthy news ecosystem, be a good partner, and create mutually beneficial opportunities. We partnered with Associated Press, Axel Springer, American Journalism Project and NYU⁠(opens in a new window). 

We also engaged in good-faith discussions with The New York Times, but on December 27, 2023, without warning, we learned of their lawsuit by reading the Times. It was a surprise and disappointment. We disagree with their claims.

Understanding copyright law and fair use

As the United States Supreme Court has explained⁠(opens in a new window), “the ‘fair use’ defense allows the public to use not only facts and ideas contained in a copyrighted work, but also expression itself in certain circumstances,” including for “purposes such as criticism, comment, news reporting, teaching…scholarship, or research.”

There are four non-exhaustive factors⁠(opens in a new window) that courts consider when determining whether something is fair use:

  1. Purpose and character of the use, including whether the use is of a commercial nature or is for nonprofit educational purpose.

  2. Nature of the copyrighted work.

  3. Amount and substantiality of the portion used in relation to the copyrighted work as a whole.

  4. Effect of the use upon the potential market for or value of the copyrighted work.

We believe generative AI training and use qualify as fair use at a minimum because they are transformative, non-expressive analytical uses that do not substitute for, or harm, the markets for the original works.

For example, our models are used in a wide range of transformative ways:

  • Scientists and mathematicians are using them to help advance their respective fields, as well as in efforts to accelerate drug development.

  • The Government of Iceland, meanwhile, partnered with OpenAI on a project aimed at using our models to help preserve the Icelandic language. 

  • Some of the world’s best-known enterprises are also finding new ways to deploy our technology across a range of sectors. Our December 2025 state of enterprise AI report highlights just some of the ways our models are delivering positive outcomes for organizations on a number of different fronts. 

  • Small business owners are also turning to AI to grow their businesses, as well as founders and entrepreneurs. GPT‑5.2 achieving a new state-of-the-art result on GDPval in late 2025—an evaluation that measures clearly defined knowledge-work tasks across 44 professions—is illustrative of the ways this technology can be valuable to entrepreneurs with access to limited resources. 

  • Importantly, AI also continues to be a valuable resource for educators. 

These use cases aren’t hypothetical—they’re observable in the real world right now. AI is already playing a tangible role in growing the broader economy, accelerating scientific discovery, improving healthcare outcomes, and making education more accessible.