U.S., China to discuss AI risks, role of government in providing safety at meeting set for Tuesday

The U.S. and Chinese governments are holding their first meeting on AI in Geneva to discuss the risks and safety concerns associated with the rapid development of AI technologies in both commercial and military sectors.

The talks aim to identify areas of concern and exchange views on the technical risks of AI, with an emphasis on advanced systems. Specific risks expected to be raised include deepfakes and “hallucinations” produced by powerful AI models.

While there were rumors of a potential agreement to ban the use of AI in certain weaponry, no such deal has yet emerged. Both countries have expressed fears about AI-related risks, with the U.S. closely monitoring the development of emerging technologies.

U.S. national security officials are particularly concerned about the potential misuse of deepfakes, especially during sensitive periods like Election Day. The intelligence community is working to incorporate AI tools while ensuring human oversight to prevent errors and hallucinations.

The Department of Defense is moving forward with adopting autonomous systems, raising questions about the appropriate level of human involvement in AI decision-making processes. While the U.S. insists that decisions involving nuclear weapons will only be made by humans, China and Russia have not made similar commitments, prompting calls for clarity on their stance.