• While dealing with the issue of AI and intellectual property rights on an international level 'would be ideal,' IP rights 'usually are territorial rights ... every country has its own system,' says University of Cambridge professor of law

• 'One, in my view, common trend which I find somewhat worrying is the very increasing and ever-continuously increasing power of major intermediary platforms,' says Ruse-Khan
  

ISTANBUL 

The rise of artificial intelligence is testing the limits of copyright law, with courts around the world grappling with whether developers can use protected content to train generative AI models without permission.

Henning Grosse Ruse-Khan, a professor of law at the University of Cambridge, says this legal gray area is at the center of mounting court battles across the US, UK, Germany, and beyond.

“Is there a so-called argument for fair use, like an exception allowing you to do this without permission?” he said in an interview with Anadolu. “That’s a key question that’s being litigated in cases in different jurisdictions … and at the moment it’s quite open.”

At the heart of the debate is the way generative AI systems — such as ChatGPT or Google Gemini — are trained. These tools rely on massive datasets, most of which include text, images, audio, and video already protected under copyright.

“The amount of data you need would imply that you have to ask basically everyone on the internet for permission, which you can't do, because how are you going to do that?" Ruse-Khan said.

While some court decisions suggest limited uses might be permitted under “fair use,” cases where AI-generated content competes directly with the originals it was trained on are more likely to be ruled unlawful, he said. A recent US federal court ruling in a case involving ROSS Intelligence found that the unauthorized use of copyrighted legal material to train AI models does not qualify as fair use.

These diverging outcomes highlight the lack of a unified global approach to intellectual property rights. “Dealing with it on an international level would be ideal,” Ruse-Khan said, adding, however, that “IP rights usually are territorial rights, so Türkiye has its IP rights system, Germany has, the US, every country has its own system.”

And that, he noted, makes global coordination difficult. “If you don't have a common position, it's very difficult to find an international consensus,” he added.

Although treaties such as the TRIPS Agreement and pacts like the Anti-Counterfeiting Trade Agreement (ACTA) aim to bring countries into alignment, Ruse-Khan said differences in enforcement and interpretation persist. “It’s difficult to have a one-fits-all solution,” he said. “It might be context-specific. Solutions which would work in Türkiye might not exactly be the solutions ideal for the United Kingdom or for South Korea or for Malawi.”

Each country, he stressed, needs some degree of legal flexibility to craft rules that fit its own specific needs.

Beyond the legal complexities, Ruse-Khan also voiced concern over the growing influence of large digital platforms—particularly their role in determining what is visible or censored online.

“One, in my view, common trend which I find somewhat worrying is the very increasing and ever-continuously increasing power of major intermediary platforms,” he said.

“The mechanisms which these platforms employ essentially are the ones which decide what is available on YouTube in terms of potential hate speech content or in terms of allegedly copyright-infringing content,” he added.

Ruse-Khan warned against assuming such corporations would prioritize the public interest. “You can’t expect these corporations essentially to operate for the public good.

“They might occasionally act in a way which promotes the public good, but in principle, they’ll be primarily interested to maximize profit. That’s what the private law regimes we have in a capitalist economy do.”