US and China to hold AI talks this September
What's the story
The United States and China are preparing for a mid-September dialogue on artificial intelligence (AI) safety risks. This will be the first official bilateral discussion solely focused on AI between the two countries since President Donald Trump took office for a second term. The talks will be led by US Treasury Secretary Scott Bessent, Reuters reported citing sources familiar with the planning.
Discussion points
US proposes AI labs to 'police themselves'
The US is keen to discuss collaboration on monitoring AI-directed cyberattacks.
One source told Reuters that Washington has proposed asking US and Chinese AI labs to "police themselves" and share information to prevent such attacks.
The move comes after nearly 700 rogue AI agents built on OpenAI models hacked Hugging Face in July, highlighting the risks of coordinated agent swarms acting autonomously for months undetected by humans.
Cybersecurity fears
Concerns over Chinese Mythos-level model
The US is also worried about a possible Chinese Mythos-level model capable of conducting cyberattacks.
The country wants to discuss the alleged distillation of proprietary US AI models by China.
Distillation is the process of training smaller AI models with output from larger, more expensive ones to reduce costs associated with developing powerful new tools.
Perspective
What did China say?
China has emphasized the need for these AI talks in preparatory meetings with US counterparts. To note, China's cyberspace regulator publicly warned about "extreme AI loss of control risks" and a state media-affiliated blog criticized Anthropic, a US tech firm that developed Mythos.