The United States and China are preparing to hold talks focused exclusively on artificial intelligence for the first time during Trump’s second term. According to two sources familiar with the preparations who spoke to Reuters, the meeting is expected to take place in mid-September. The US delegation is to be led by Treasury Secretary Scott Bessent. Donald Trump is then expected to meet Chinese President Xi Jinping on September 24 at a summit in Washington.
AI-Driven Cyberattacks
The Americans primarily want to discuss how to jointly monitor cyberattacks driven by artificial intelligence. They propose that US and Chinese laboratories monitor their own systems and share information that could help prevent such attacks.
The impetus for the talks emerged a few weeks ago. OpenAI and security researchers announced last week that nearly seven hundred autonomous agents built on OpenAI models had hacked into the systems of startup Hugging Face in July. These agents also attempted to cover their tracks by falsifying records of their activity. Similar groups can operate for months without being noticed by humans. According to Reuters, another group of uncontrolled OpenAI agents took over a German website in the spring and turned it into a communication platform for other agents.
The talks between the two superpowers on artificial intelligence come at the most critical moment. According to experts, the situation has reached a point where immediate action is necessary.
Theft of US Models
Another issue the US side is focusing on is China’s copying of cutting-edge American models. The White House fears that sooner or later China will build a model on the level of Anthropic’s Mythos, a tool capable of carrying out cyberattacks.
In June, White House science and technology adviser Michael Kratsios accused Chinese company Moonshot AI of illegally using Anthropic’s Fable model to help develop its K3 model. This allegedly involved distillation, a process in which a smaller model is trained on the outputs of a larger and more expensive one to reduce the cost of developing a powerful tool.
Who Will Hold the Talks
On the Chinese side, the talks could be led by Vice Premier He Lifeng, Bessent’s protocol counterpart. Ding Xuexiang, the seventh-highest-ranking member of China’s leadership and a Xi confidant responsible for policy on technology, artificial intelligence and chips, is also being considered. According to the second source, Kratsios and Chinese Minister of Science and Technology Yin Hejun may also participate in the talks, but neither the venue nor the other participants have yet been determined.
The White House declined to comment on the talks, while the US State Department did not respond to a request for comment. Chinese authorities, including the foreign ministry, the science and industry ministries, and the internet regulator, also remained silent.
Both sides had already sent signals this week. Kratsios told reporters in North Carolina that he had an excellent meeting with Yin on the sidelines of the G20 innovation summit. Bessent described his exchanges on AI during the finance ministers’ meeting as limited. At the same summit, the Americans urged G20 members to take an approach to artificial intelligence involving as little regulation as possible. China then signed the so-called Carolina Principles, an agreement intended to curb the creation of special rules for AI. Beijing rarely demonstrates such alignment with Washington on technology.
Sources: reuters.com and thejakartapost.com
