US raises concerns to Chinese officials about AI misuse

WASHINGTON (Reuters) -U.S. officials flagged concerns over China's "misuse" of artificial intelligence in their first formal bilateral talks on the issue, the White House said on Wednesday, as the superpowers seek to avoid confrontation over the fast-developing technology.

During the meetings in Geneva on Tuesday, the U.S. delegation also stressed to their Chinese counterparts the need to "maintain open lines of communication on AI risk and safety as an important part of responsibly managing competition," the White House said.

"The United States also raised concerns over the misuse of AI, including by (the People's Republic of China)," White House National Security Council spokesperson Adrienne Watson added.

Worries are mounting among U.S. officials about China's access to AI technology, amid fears that it could be used by Beijing to upend elections in other countries, create bioweapons and launch cyberattacks.

The State Department already has pressed China and Russia to match U.S. declarations that only people, and never artificial intelligence, would make decisions on deploying nuclear weapons.

Reuters reported this month that President Joe Biden's administration is poised to open up a new front in its effort to safeguard U.S. AI from China and Russia with preliminary plans to place guardrails around the most advanced AI models, the core software of artificial intelligence systems like ChatGPT, sources said.

A bipartisan group of U.S. lawmakers unveiled a bill last week that would make it easier for the Biden administration to impose export controls on AI models. A bipartisan group of U.S. senators, including Majority Leader Chuck Schumer, on Wednesday called for a jump in government research funding of artificial intelligence as they debate new legal safeguards.

The United States has taken measures to stem the flow to China of American AI chips and the tools to make them. The Biden administration also proposed a rule to require U.S. cloud companies to tell the government when foreign customers use their services to train powerful AI models that could be used for cyber attacks.

China has been heavily relying on many open source models developed in the West such as the Meta Platforms "Llama" series and many advanced American AI chips are making their way into China.

(Reporting by Alexandra Alper and Jasper Ward; writing by Paul Grant; Editing by Caitlin Webber and Will Dunham)