In this episode of the High Capacity podcast, I talk with Matt Sheehan, Senior Fellow at the Carnegie Endowment, about the next phase of the US-China AI race: AI safety, open-source models, AI agents, job displacement, and what the US and China might discuss with each other. As the U.S. and China prepare for formal talks on AI, both sides are wrestling with a fast-changing technology landscape. The U.S. is increasingly worried about frontier model risks, cyber threats, non-state actors, and pre-release testing. China, meanwhile, is becoming more concerned about AI’s impact on jobs, social stability, political control, and the rise of autonomous AI agents. We discuss what China really fears about AI, why Beijing’s AI governance strategy is shifting from control to deployment to broader regulation, how open-source AI might complicate safety talks, and what realistic US-China talks on AI might look like. Topics covered: U.S.-China AI talks, AI safety, Chinese AI policy, AI agents, OpenClaw, open-source AI, frontier models, cyber risk, AI job displacement, China’s AI regulations, autonomous weapons, AI hotlines, and the future of AI governance. Links: - Matt Sheehan’s Substack - Matt Sheehan’s Carnegie page - “China is getting worried about AI & jobs” - “China Is Worried About AI Companions. Here’s What It’s Doing About Them.” - Matt Sheehan on Twitter: @mattsheehan88 For transcripts and more, check out the High Capacity newsletter More about the host, Kyle Chan, fellow at Brookings
