Bessent prepares for talks with China on AI — TIME
U.S. Treasury Secretary Scott Bessent is preparing to meet this weekend with Chinese Vice Premier He Lifeng to discuss artificial intelligence. As TIME writes, the meeting is being prepared amid distrust between the United States and China and the rapid development of advanced AI systems.
U.S. President Donald Trump previously called public warnings about the uncontrolled development of artificial intelligence a hoax. At the same time, the Chinese state-run outlet China Daily said that leading U.S. AI companies are using calls to slow technological development to restrain China’s progress in this field.
Technical exchanges on AI safety
The author of the TIME article suggests that Washington and Beijing should not try to immediately conclude a large-scale agreement to restrict the development of advanced models. Instead, the countries should begin technical exchanges on testing systems for dangerous capabilities and behavior.
The article names among the risks AI’s ability to help untrained people create biological weapons, as well as to evade human oversight and control. U.S. AI companies have been studying such threats for several years, while corresponding safety-testing requirements have also begun to appear in Chinese technical standards.
More current news is available on the UA.News Telegram channel Telegram.
According to the author, exchanging practices for testing, monitoring, and deploying safeguards could improve the safety of models in both countries. Such cooperation would not require complete trust or an identical interpretation of risks, since both sides are interested in maintaining control over powerful systems regardless of where they are developed or used.
Protocol for crisis incidents
Another possible step, the author says, is the creation of an intergovernmental communication protocol in the event of AI-related incidents. Examples include situations in which a U.S. model gets out of a company’s control and hacks the networks of a Chinese company or the Chinese government, or when a Chinese AI drone attacks a U.S. Navy ship and the cause of the incident cannot be quickly established.
For such a channel, it is necessary to define incidents that must be reported, responsible contact persons, and the procedure for notifying about unauthorized actions without disclosing sensitive intelligence data. The author also proposes transmitting initial information in writing rather than only by phone in order to reduce the risk of misunderstanding technical details.
TIME emphasizes that dialogue on AI safety should be separated from disputes over U.S. export controls on advanced chips and equipment for their production. The goal of such contacts should not be to end technological competition, but to prevent consequences undesirable for both countries.
