Get Alex Colville on Dwarkesh Podcast

Alex Colville has been breaking timely stories on how China is exporting its AI models and governance to the world, including a December 12, 2025 investigation into Guangxi’s push across ASEAN and a Malaysian government backed “shariah aligned” chatbot built with China’s DeepSeek, raising hard questions about values, bias, and influence in Southeast Asia. He followed up on December 18, 2025 with original testing that jailbroke DeepSeek-R1 and Alibaba’s Qwen-3-Max to produce step by step instructions for fentanyl and explosives, undermining claims that China should lead global AI safety.

Dwarkesh’s audience is wrestling right now with who sets AI rules and whose models shape the Global South, and Colville has the receipts and recent fieldwork to answer with specifics. Having him on today means interrogating the World AI Cooperation Organization push, export strategies running through provincial hubs like Guangxi, and the gap between Beijing’s rhetoric and real model behavior that could affect U.S. policy and industry within months.

If you want a no spin briefing on the geopolitics of AI that actually cites fresh tests and documents, back this booking now.