News industry urges courts to defend copyright against AI companies

By Danielle CoffeyPresident and CEO of the News/Media AllianceFrom our country’s beginning, the Founders recognized that American democracy would rely upon a flourishing independent press. The Constitution protects a free press through the First Amendment, of course, but also through a less celebrated provision: its recognition of the need for copyrights. Copyright gives creators the economic incentive to develop and disseminate their ideas, and the authors of the Constitution intended copyright, the late Justice Sandra Day O’Connor explained, to be “an engine of free expression” that would power an independent press.But that engine is now in jeopardy. AI companies are routinely stealing copyrighted content with impunity. The courts should seize the opportunity before them to put a stop to this abuse and reaffirm the Founders’ intentions.In 2023, the New York Times sued Microsoft and OpenAI, alleging that OpenAI trained its AI models on the newspaper’s content without permission, and that OpenAI and Microsoft copied real-time content for use in outputs. Other news publishers followed suit, all alleging that OpenAI and Microsoft’s AI products directly and unfairly compete with their journalism. Over a dozen of the earliest suits, including The New York Times v. Microsoft, are now at the summary judgment stage, where a judge will soon rule for the first time on the legality of AI tools taking copyrighted news content for real-time answers. This is a pivotal moment for the news media industry and the intellectual property laws that have made our country’s economy so strong. That’s why the News/Media Alliance, the association I lead that represents the entire news media industry, just filed an amicus brief telling the court that for the sake of the free press and all that it means to our country, it’s critical they rule in favor of the news publishers. What OpenAI and Microsoft are doing does more than harm individual publishers. They are undermining a bargain that has for years linked search, traffic, and content production on the open web. Traditional search engines direct users to the sources that invest in discovering news. But AI, and ChatGPT specifically, provide real-time answers - either summaries or verbatim copies of the news - that substitute for visiting news sites. Studies show that when individuals start using AI chatbots, they unsurprisingly use traditional search engines and go to news sites much less. This drives down publishers’ subscription and advertising revenue by using their own work product against them.By law, using copyrighted content in this manner, without permission or compensation, is illegal. But OpenAI and Microsoft think they have found a way to get around copyright laws by arguing that the way they use news content is “fair use.” If you’re parodying, criticizing, or commenting on news content, you typically don’t have to compensate the people who produced that content. Newspapers are major proponents of fair use - news reporting is one of the prime examples mentioned in the Copyright Act as fair use. But what AI companies are doing goes too far. None of the legal precedents support stealing content to compete with its producers as legitimate cases of fair use.OpenAI and Microsoft try to defend their conduct by noting that they sometimes repackage the content they are taking instead of literally copying it (although they do literally copy it often). The courts have long understood that summarizing or excerpting copyrighted material can violate the creator’s rights if it serves a substitutional purpose. There are few things more obviously substitutional for a news consumer than getting a question answered with a ChatGPT query rather than reading a news article that information came from in the first place.If OpenAI and Microsoft get their way, the subsequent weakening of copyright law would pose an existential threat to the news industry, ultimately depriving millions of Americans of news about their country or their communities. We’re never going to go back to a world where consumers can’t use AI, nor would we want to. But it’s critical that news publishers are properly compensated for the use of their content. AI companies don’t expect utilities to power their data centers for free, and they don’t expect to be able to obtain chips for free. So why would they expect to be able to acquire the information that powers their models for free?There is already a robust licensing market for news publishers to sell their content to AI companies. These licensing deals offer a sustainable, cooperative path forward for both industries. The AI industry gets the information it needs, and news publishers get the revenue they need to continue producing quality content. But it all hinges on publisher control of content. If companies like OpenAI are allowed to steal news content, everything falls apart, and millions of Americans will lose access to high-quality journalism.