Zuckerberg: US Should Compete, Not Block Chinese AI Models

Meta CEO Mark Zuckerberg is raising questions about whether American AI labs truly want Chinese AI models restricted in the US. He believes banning them is not an effective solution. Zuckerberg suggests the US should compete by addressing bottlenecks, rather than blocking rivals. He also warns against "regulatory capture" by American labs like OpenAI and Anthropic, implying they might use policy to limit competition. He noted mixed signals about these frontier labs truly wanting open-source models to succeed. These comments follow the release of Moonsht AI's Kimi K3 model, which is nearing the capabilities of leading US models. Zuckerberg supports open models that users can customize, arguing that restrictions on powerful AI tools are not the answer and open models can enhance cybersecurity. His intervention highlights a growing division among US tech leaders on how to manage Chinese competition and regulate AI. This matters because it reveals fundamental disagreements on the future direction and control of artificial intelligence.