Switzerland’s parliament is debating a law that would shield the nation’s newspapers, magazines, broadcasters and other media companies from AI data harvesting.
Keystone / Gaetan Bally
Legislators and courts around the world are scrambling to apply and update copyright protection laws to face the threat posed by artificial intelligence (AI).
+Get the most important news from Switzerland in your inbox
Switzerland’s parliament is debating a law that would shield the nation’s newspapers, magazines, broadcasters and other media companies from AI data harvesting. Lawmakers are trying to balance the need to protect existing copyright with the potential to build new digital systems that can bring benefits to society.
Many other countries are grappling with the same conundrum and are exploring a patchwork of solutions. This is being monitored by the Geneva-based World Intellectual Property Organization (WIPO). The United Nations agency acts as a forum for 194 member states to find consensus on intellectual property rules.
+ Switzerland considers AI data ‘theft’ defences
WIPO says it “serves the world’s innovators and creators, ensuring that their ideas travel safely to the market and improve lives everywhere”. Copyright rules must also make room for technological innovation – which sometimes results in conflict with copyright holders.
Lawsuits take aim at AI companies, mainly in US
“Copyright has been challenged by many forms of technological disruption in the past,” Garrett Levin, Senior IP Law and Policy Officer on IP and Frontier Technologies, tells Swissinfo.
“But AI can produce content indistinguishable from human-generated content by training on copyrighted material. That is what sets AI apart from previous disruptions.”
This combination of training on copyrighted data and the creation of new, human-like works sets AI apart from other disruptions, such as photography, photocopying and, more recently, YouTube and music streaming. A 2024 report by the International Confederation of Societies of Authors and Composers (CISAC) estimated that AI could result in musicians’ revenues being reduced by a quarter by 2028.
The AI juggernaut has resulted in numerous lawsuits claiming copyright infringement over works that were used to train AI without being licensed or paid for. According to the website chatgptiseatingtheworldExternal link, which tracks ongoing legal cases, there were 79 active copyright lawsuits against AI companies on October 17 of this year. Two-thirds of these cases had been initiated in the US.
More
More
Artificial intelligence explained
Only a handful of lawsuits originated outside of North America. For example, OpenAI, the creator of ChatGPT, is being sued by the Folha de São Paulo newspaper in Brazil and India’s Asian News International.
The Japanese Yomiuri Shimbun, Nikkei and Asahi Shimbun media group are seeking damages against the AI company Perplexity.
‘Copyright law can address these challenges’
The most common counter argument from AI companies is that of “fair use”— a doctrine that’s often used to support new work based on existing content, such as criticism, scholarship, parody, and adaptation. It is unclear which side will prevail. Most lawsuits are ongoing, and the handful of court verdicts have produced mixed results.
For example, the German music royalty collection agency GEMA this month won a lawsuit lodged in a Munich court against OpenAI. But the US stock photo company Getty Images lost most of its arguments in a case brought against Stability AI in London.
+ Switzerland aspires to build ‘human’ artificial intelligence
Levin believes the plethora of lawsuits is a sign that copyright law is robust and functioning properly, with courts adjusting to new applications of existing rules. “Rights holders are not calling for massive changes to copyright law, they are availing themselves of existing copyright law,” he says.
“It is far too early to say that we must tear everything up and re-write copyright law. Even at the speed that AI is evolving, I am confident that copyright law is capable of addressing these challenges.”
Switzerland alone in debating AI and copyright
Other disputes have been settled out of court with licensing deals that involve AI companies paying royalties to content producers.
The US AI company Anthropic has agreed to pay $1.5 billion to book authors whose material was used to train AI models. The New York Times, Washington Post, Axel Springer and music labels Warner, Universal and Sony have also struck deals with AI firms.
Some countries are addressing the impact of AI on intellectual property through legislation, but the Swiss parliament is alone in actively debating a law tailored to this specific problem, according to Levin.
“Most legislation is directed at creating national strategies to tackle broad range of policy questions, such as data privacy, IP and human rights,” he said. This approach is being adopted by the likes of EU, Brazil and South Korea.
International treaty unlikely
AI technology is still in its infancy, and the scope of its eventual impact is still unknown. Regulators are playing catch up, and countries are generally content to pursue their own legal remedies.
“There is always a risk that disjointed policy making around the world challenges the bridge to consensus,” Levin says. “But we have not yet seen much appetite among member states for an international treaty around AI and IP.”
“We need to look long and hard at the balance between incentivising human creativity and encouraging the development of revolutionary technology,” he adds. “We must ensure that both can exist side by side.”
Edited by Gabe Bullard/VdV
About the Author
