Openai Has Little Legal Recourse Against Deepseek, Tech Law Experts Say

OpenAI has limited legal options if it wants to take DeepSeek to court.

picture alliance/dpa/Getty Images

• OpenAI and the White House have accused DeepSeek of using ChatGPT to cheaply train its new chatbot.
• Experts in tech law say OpenAI has little recourse under intellectual property and contract law.
• OpenAI's terms of use may apply, but are largely unenforcible, experts say.

This week, OpenAI and the White House accused DeepSeek of something akin to theft.

In a flurry of press statements, they said the China-based upstart had bombarded OpenAI's chatbots with queries, hoovering up the resulting data trove to quickly and cheaply train a model that's now almost as good.

The Trump administration's top AI "czar" said this training process, called "distilling," amounts to intellectual property theft. OpenAI, meanwhile, told Business Insider and other outlets that it is investigating whether "DeepSeek may have inappropriately distilled our models."

OpenAI is not saying if the company plans to pursue legal action, instead promising what a spokesperson termed "aggressive, proactive countermeasures to protect our technology."

But could they? Could they sue DeepSeek on "you stole our content" grounds, much like the grounds OpenAI was itself sued on in an ongoing 2023 copyright claim filed by The New York Times and other news outlets?

Business Insider posed this question to experts in technology law, who said challenging DeepSeek in the courts would be an uphill battle for OpenAI, now that the content-appropriation shoe is on the other foot.

OpenAI would have a hard time proving an intellectual property or copyright claim, these lawyers said.

"The question is whether ChatGPT outputs" — meaning the answers it generates in response to queries — "are copyrightable at all," said Mason Kortz of Harvard Law School.

That's because it's unclear that the answers ChatGPT spits out qualify as "creativity," he said.

"There's a doctrine that says creative expression is copyrightable, but facts and ideas are not," explained Kortz, who teaches at Harvard's Cyberlaw Clinic.

"There's a huge question in intellectual property law right now about whether the outputs of a generative AI can ever constitute creative expression or if they are necessarily unprotected facts."

Could OpenAI roll those dice anyway, and claim that its outputs actually are protected?

That would be unlikely, the lawyers said.

OpenAI is already on the record in the New York Times copyright case arguing that training AI is an allowable "fair use" exception to copyright protection.

If they do a 180 and tell DeepSeek that training is not a fair use, "that might come back to kind of bite them," said Kortz. "DeepSeek could say, 'Hey, weren't you just saying that training is fair use?'"

There's arguably a distinction between the Times and DeepSeek cases, Kortz adds.

"Maybe it's more transformative to turn news articles into a model" — as the Times accuses OpenAI of doing — "than it is to turn outputs of a model into another model" as DeepSeek may have done, Kortz said.

"But this still puts OpenAI in a pretty tricky situation with regard to the line it's been towing regarding fair use."

A breach of contract lawsuit is more likely

A breach-of-contract lawsuit is much likelier than an IP-based lawsuit, though it comes with its own set of problems, said Anupam Chander, who teaches technology law at Georgetown University.

The terms of service for Big Tech chatbots like those developed by OpenAI and Anthopic forbid using their content as training fodder for a competing AI model.

"So perhaps that's the lawsuit you might possibly bring — a contract-based claim, not an IP-based claim," Chander said.

"Not 'you copied something from me,' but that you benefited from my model to do something that you were not allowed to do under our contract."

There's a possible hitch, Chander and Kortz say. OpenAI's terms of service require that most claims be resolved through arbitration, not lawsuits. There's an exception for lawsuits "to stop unauthorized use or abuse of the Service or intellectual property infringement or misappropriation.

There's a larger hitch, though, experts say.

"You should know that the brilliant scholar Mark Lemley and a coauthor argue that AI terms of use are likely unenforceable," Chander said. He was referring to a January 10 paper, The Mirage of Artificial Intelligence Terms of Use Restrictions, by Stanford Law's Mark A. Lemley and Peter Henderson of Princeton University's Center for Information Technology.

To date, "no model creator has actually tried to enforce these terms with monetary penalties or injunctive relief," the paper says.

"This is likely for good reason: we think that the legal enforceability of these licenses is questionable," it says. That's in part because model outputs "are largely not copyrightable" and because laws like the Digital Millennium Copyright Act and the Computer Fraud and Abuse Act "offer limited recourse," it argues.

"I think they are likely unenforceable," Lemley told BI of OpenAI's terms of service, "because DeepSeek didn't take anything copyrighted by OpenAI, and because courts generally won't enforce agreements not to compete in the absence of an IP right that would prevent that competition."

Lawsuits between parties in different nations, each with its own legal and enforcement systems, are always tricky, Kortz said.

Even if OpenAI cleared all the above hurdles and won a judgment from a US court or arbitrator, "in order to get DeepSeek to turn over money or stop doing what it's doing, the enforcement would come down to the Chinese legal system," he said.

Here, OpenAI would be at the mercy of another extremely complicated area of law — the enforcement of foreign judgments and the balancing of individual and corporate rights and national sovereignty — that stretches back to before the founding of the United States.

"So this is, a long, complicated, fraught process," Kortz added.

Could OpenAI have protected itself better from a distilling incursion?

"They could have used technical measures to block repeated access to their site," Lemley said. "But doing so would also interfere with normal customers."

He added, "I don't think they could, or should, have a valid legal claim against the searching of uncopyrightable information from a public site."

Representatives for DeepSeek did not immediately respond to a request for comment.

"We know that groups in the PRC are actively working to use methods, including what's known as distillation, to try to replicate advanced U.S. AI models," OpenAI spokesperson Rhianna Donaldson told BI in an emailed statement.

"We are aware of and reviewing indications that DeepSeek may have inappropriately distilled our models, and will share information as we know more," the statement said. "We take aggressive, proactive countermeasures to protect our technology and will continue working closely with the US government to protect the most capable models being built here."