Zhipu AI Founder Defends Open-Source Models Amid Security Debate

The founder of Chinese AI startup Zhipu is making a forceful case for keeping frontier artificial intelligence models openly accessible, even as governments and competitors move toward tighter restrictions on advanced systems.

Tang Jie, who leads Zhipu while teaching at Tsinghua University, contends that genuine AI safety emerges from transparency, broad developer participation, and public oversight—not from limiting who can access cutting-edge models. Zhipu recently acted on that philosophy by releasing its GLM-5.2 model under an open-source license that permits downloading, modification, and commercial use, Bloomberg first reported.

Diverging approaches to frontier AI

The stance puts Zhipu at odds with a growing movement toward restriction. Anthropic has limited access to some of its most capable systems citing national security considerations. Reuters has reported that Beijing itself is weighing constraints on overseas access to certain Chinese-developed AI models.

Tang's position reflects a broader pattern in China's AI sector, where open-source development has become the dominant strategy. Major players including Alibaba have released model families like Qwen publicly, helping Chinese developers close the technology gap with U.S. competitors and establish themselves as significant contributors to the global open-source AI ecosystem.

Investment over monetization

Zhipu is charting an unusual course by prioritizing technological advancement over near-term revenue. Tang said the company will focus the next two years on developing capabilities in long-horizon reasoning, autonomous AI agents, and self-training models rather than aggressively commercializing applications.

The GLM-5 platform targets complex coding and agentic AI tasks, with performance benchmarked against Anthropic's Claude Opus series. Zhipu recently announced a $4 billion share sale in Hong Kong and disclosed plans for a Shanghai listing, signaling strong investor confidence in China's expanding AI market.

The security tension

The debate over access controls has intensified as frontier models demonstrate increasingly sophisticated capabilities. Recent systems have shown they can identify complex software vulnerabilities with minimal human guidance, prompting both governments and AI companies to strengthen safeguards around their most advanced offerings.

Why it matters

This disagreement over open versus restricted AI development will shape the technology's trajectory and competitive landscape. If major Chinese developers maintain open-source strategies while Western companies restrict access, it could accelerate global adoption of Chinese models and shift influence in the AI ecosystem. For enterprises, the availability of capable open-source alternatives affects vendor lock-in risk, customization options, and total cost of ownership for AI infrastructure.

Bloomberg first reported the details of Tang's comments and Zhipu's strategic direction.