TheChek News Flow

AI executives loved it when President Trump sided with them against copyright holders—but the White House dodged the topic in its AI Action Plan. Here’s why.

Before the White House released its AI Action Plan in July, people involved with drafting the document weighed whether to include recommendations that could have impacted the contentious relationship between artificial intelligence companies and copyright holders.

Publishers, authors, film and music companies have filed more than 40 lawsuits in the past few years, arguing that AI firms have infringed on their copyrights by using their data without permission to train large language models. AI companies have argued they’re abiding by fair use, a legal doctrine that allows limited use of copyrighted material without permission from rights holders in certain situations.

Any move by President Donald Trump’s administration to take a side in the dispute could undercut the efforts of publishers and Hollywood to protect their copyrights. In the months leading up to the AI Action Plan’s release, groups representing the music, film and media industries got wind that the document would address copyright in some way, according to a person in the music industry.

In meetings with different government agencies involved in developing the plan, as well as key figures including David Sacks, White House AI czar, and Michael Kratsios, White House science and tech adviser, representatives of publishers, music and film companies argued that the debate about copyright should be left to the courts, the person said.

When the plan came out on July 23, it made no mention of copyright. That was because the people who put the plan together couldn’t figure out what kind of policy recommendation they could actually make, according to a person with knowledge of the situation.

So it was a surprise to audiences at the Winning the AI Race summit—an event in Washington hosted later that day by the “All-In” podcast—when President Donald Trump weighed in on the issue on the side of the AI firms. “You can’t be expected to have a successful AI program when every single article, book or anything else that you’ve read or studied, you’re supposed to pay for,” he said.

AI executives were elated by the comments. One of them said they hoped Trump’s words would influence court decisions or prompt government bodies to take a similar stance. A music industry official predicted the courts won’t care about those comments, however.

Legally, Trump’s words have little significance. A patchwork of state and federal laws protects intellectual property in the U.S., and many of those regulations, including the Copyright Act, have been around for decades. Any changes to federal law would need to go through Congress.

Tod Cohen, a partner at Manatt, Phelps & Phillips in San Francisco who previously worked at Twitter and eBay, said the administration would have the option of filing an amicus brief in an AI copyright lawsuit, outlining its position. Such briefs can be persuasive but are not legally binding.

Cohen said the challenge for the administration if it wanted to set a more general rule around fair-use disputes is that each situation has a specific set of facts and requires evaluation on a case-by-case basis. “If you wrote, ‘The position of the government in the AI Action Plan is training data is fair use,’ that doesn’t help a court at all to help determine what the law is, because you have to evaluate it by what the facts are that are presented in each case,” he said.

To Cohen, the companies’ stance in the broader debate boils down to money. “‘We can get away with not paying’ is the ultimate position of them all,” he said.

Much is at stake. The New York Times’ case against OpenAI is heading toward the deposition phase. Courts in two other copyright cases ruled partly in favor of Anthropic and Meta. But the judge in the Anthropic case also ruled in July that the book authors involved in the dispute could move forward with a class action suit against Anthropic, potentially exposing the company to billions of dollars in penalties.

Some AI firms have struck licensing deals with publishers. OpenAI has signed numerous deals with publishers such as Vox Media and News Corporation, allowing the ChatGPT creator to use their data for training its models, in exchange for payment of some kind. Damon Beres, a senior editor at The Atlantic—one of the publishers who reached a deal with OpenAI—wrote an essay calling it “a devil’s bargain.”

Others, such as Google, have not made such deals, preferring to rely on the fair use legal doctrine to protect their use of publishers’ data in the training of Google models.

The disconnect between the president’s public position and the plan itself shows how unpredictable Trump can be on issues of the day—and it illuminates inconsistencies in the government’s approach to AI. In the same speech in Washington, Trump criticized state AI regulation and advocated for a federal rule, just weeks after the Senate killed an industry-backed effort to ban states from regulating AI for a decade.

The White House and the Office of Science and Technology Policy did not respond to requests for comment.

The situation also illuminates the growing role of technology companies in pushing AI policy. The CEOs of Silicon Valley companies such as Meta Platforms, Nvidia and OpenAI have all been trying hard to earn Trump’s favor since he was elected, making frequent trips to his private Mar-a-Lago club. The same companies have also been pouring millions of dollars into federal lobbying efforts, seeking to capitalize on a business-friendly administration that is open to rolling back regulations for AI companies if it means beating China in the AI race.

Meanwhile, the key White House tech advisers behind the AI plan, including Sacks and Kratsios, have close ties to Silicon Valley.