U.S. and China open high-level talks on AI use and risks

Top officials from Washington and Beijing meet for the first time in Geneva to share their mutual concerns about each other’s use of artificial technologies (AI). It’s hoped that the meetings will lead to further dialogue to ensure the technology is not misused by either party.

On 14 May 2024, top officials and envoys from the U.S. and China met in Geneva, Switzerland, to open the first-ever bilateral dialogue between the two nations to ensure that artificial intelligence (AI) technologies are not misused or become an existential risk. The talks were initially agreed by Presidents Joe Biden and Xi Jinping when they met on 15 November 2023 at the Filoli Estate near San Francisco.

The U.S. and China express concerns about the use of AI

While no concrete results were expected, it did present the opportunity for both sides to express their concerns regarding each other’s use of AI. The meeting was led by high-level officials – who remained anonymous – from the White House and Departments of State and Commerce. At the same time, U.S. officials hoped that it would provide insight into Beijing’s thinking about AI amid a generally secretive Chinese approach to the technology. 

“To be very clear, talks with Beijing are not focused on promoting any form of technical collaboration or cooperating on frontier research in any matter. And our technology protection policies are not up for negotiation,” one official said in a series of statements released by both sides during and after the talks.

The U.S. delegation stressed the need to “maintain open lines of communication on AI risk and safety as an important part of responsibly managing competition.”

Furthermore, U.S. officials raised concerns about China’s “misuse of AI”, citing China’s rapid proliferation of AI capabilities across civilian, military, and national security sectors, which it feared could undermine the security of the U.S. and its allies. For example, China has built one of the world’s most intrusive digital surveillance systems, which has an AI component with street cameras and the ability to track chat apps and mobile phone use.

Banning the use of autonomous nuclear weapon technology

But one of the main areas of concern was the use of AI and nuclear weapons – on 26 April 2023, the U.S. introduced the Block Nuclear Launch by Autonomous Artificial Intelligence Act “To prohibit the use of Federal funds to launch a nuclear weapon using an autonomous weapons system that is not subject to meaningful human control, and for other purposes.”^[1]

Washington has pressed China and Russia to do the same, but so far neither has matched the U.S. pledge.

Another concern expressed by U.S. officials was the use of deepfake technology that could influence political opinions and spread misinformation campaigns. However, China – unlike the U.S. – imposed a comprehensive set of new laws in January 2023 that ban the use of manipulative AI fakery.

After a year-long public comment period, the Cyberspace Administration of China implemented what was described as the ‘first-of-its-kind’ and the ‘most comprehensive’ law in the world regulating deepfakes. The regulations prohibit the use of deepfake technology for spreading ‘fake’ news or information that could disrupt the economy or national security, with Beijing describing it at the time as a mechanism for social stability.

China expresses concerns over U.S. restrictions on AI technologies

Conversely, Beijing’s representatives took a firm stance against Washington over its “restrictions and pressure” on AI. In August 2023, Biden signed an executive order banning Chinese access to advanced AI semiconductors and U.S. investment in China’s AI sector – Beijing has lashed out against the order multiple times. China also advocated for the United Nations to take a lead role in the global governance of AI in an attempt to sidestep Washington’s influence over the technology.

On 8 May 2024, a bipartisan group of U.S. lawmakers unveiled a bill that would make it easier for Washington to impose export controls on AI models and would give the Commerce Department authority to bar American companies and individuals from working with foreign actors to develop AI systems that might pose a risk to U.S. national security. If passed, the “Enhancing National Frameworks for Overseas Critical Exports Act” (ENFORCE Act) would also prevent future AI export regulations from legal challenges.^[2]

Pamir considers that both parties are fully aware of the opportunities AI offers, but that it also poses an equal risk to both the U.S. and China. Both parties are aware that everyone has a lot to lose if AI is weaponized or misused, and it is essential that constructive talks between Washington and Beijing are maintained to ensure its prudent use, while ensuring safe competition. But the success of this first round of talks will ultimately be dictated by whether this dialogue continues and that there are further on-going discussions.

^[1] https://www.markey.senate.gov/imo/media/doc/block_nuclear_launch_by_autonomous_ai_act_-_042623pdf.pdf

^[2] https://www.govinfo.gov/app/details/BILLS-118hr8315ih