U.S. Treasury Secretary Scott Bessent announced plans for an AI safety alert system, aiming to create an open channel for both countries to send alerts about potential threats from their AI systems. The proposal, made before U.S. President Donald Trump and Chinese President Xi Jinping are set to meet, would allow the countries to share their 'vision of common goals.'
The plan involves setting up a new AI safety notification mechanism, which Bessent said could enable the countries to monitor and address risks from their AI systems.
However, the proposal comes at a time when China may not be ready to trust the U.S. on AI safety discussions, given the ongoing tensions over export controls and allegations of intellectual property theft.
China has yet to publicly acknowledge the proposed alert system, raising concerns among experts about its effectiveness. People familiar with Bessent’s plan told Politico that the system appears superficial, as it lacks officials with technical expertise to assess the urgency of risks described in communications.
"The scientific, technical term is BS," said one source, mocking the plan's lack of substance. Sam Bresnick, a research fellow at the Center for Security and Emerging Technology, noted that while the existing channel is better than creating a new one, it lacks the technical expertise needed for high-level risk assessments.
Low expectations for Trump talks were expressed by Su Lian Jye, a chief analyst at Omdia, who suggested that the AI alerts may help promote more transparency but won't reduce mutual suspicions. He emphasized that much more cooperation would be needed to make the safety mechanism meaningful.
China's suspicion of U.S. intentions in AI safety talks was highlighted by experts, with concerns that the U.S. may be using safety discussions as a way to maintain its advantage. Bresnick warned that if China feels 'AI safety' is a code word for suppressing Chinese AI, the alert system may become ineffective in the long term.
Source: arstechnica
