Zuckerberg Urges US to Ease Open-Source AI Limits to Compete with China

Meta CEO Mark Zuckerberg recently urged the US government to ease restrictions on open-source AI models to better compete with Chinese rivals. He believes AI should be accessible to all, not controlled by a few, and criticized the idea that only extreme centralization ensures safety.

Meta also launched Muse Glimmer, an AI model using 4-bit quantization that runs locally on Mac or PC with 24GB or 32GB VRAM. His remarks sparked discussions in the open-source AI community, especially as costs rise and security incidents at Anthropic and OpenAI draw attention to open-weight models.

Open-weight models generally cost less to train than closed ones and offer public core components for customization. Chinese firms are leading in this space, with Moonshot's Kimi K3, Alibaba's Qwen3.8-Max, and DeepSeek's V4-Flash competing head-to-head with top US models.