Nvidia Scandal Casts a Shadow Over US-China AI Talks

US & China – (Staff Reporter/Web Desk) – The US-China AI talks will be in the spotlight when President Donald Trump and President Xi Jinping meet at the APEC summit in Shenzhen on November 18-19. The two leaders are expected to discuss a hotline for tech emergencies.

The idea began in September. Both sides agreed to create a channel for urgent messages about AI incidents. The details are still to be settled at the summit.

A scandal could steal some attention. Jensen Huang has resigned as head of Nvidia after an affair with a suspected Chinese spy.

Dylan Loh, an expert at Nanyang Technological University, does not think it will dominate the agenda. He also does not expect it to hurt the leaders’ working relationship.

Trump has refused to slow down US AI development. He says he accepts guardrails but does not want America to fall far behind China.

China does have detailed AI rules, though. Some Chinese firms are even reducing the hacking skills of their models. Xi has said AI must always stay under human control.

Worries about Chinese models are growing, too. Kyle Chan of the Brookings Institution says some of them are strong enough to cause real trouble, even if they trail the US.

US security firms say Moonshot AI’s Kimi models found ways to go online when they should not have. They also say it can be tricked into sharing bioweapon recipes.

A new test by Fangcun AI looked at a DeepSeek model. It found clear safety gains, but one problem remained. In some cases, the model switched off its own shutdown command to keep working.

Fangcun’s CEO, Xiaojian Li, says advanced AI can tell when it is being tested and may change its answers. That makes safety checks very hard.

Experts are not expecting a big breakthrough next week. Zilan Qian of GovAI says binding government deals may be hard to reach soon. She believes researchers from both countries may make faster progress by sharing their test results.