Humanoid robots compete in the 100-meter race during the 2nd World Humanoid Robot Games at the National Speed Skating Oval in Beijing on August 25, 2026.
Wang Zhao | Afp | Getty Images
BEIJING — As US AI pioneers voice concerns about the risks of artificial intelligence, their Chinese counterparts remain largely silent.
Chinese firms like Z.ai, Moonshot, MiniMax, Alibaba, and Tencent have not echoed the warnings issued by Silicon Valley executives, nor responded to inquiries from CNBC regarding recent calls for caution.
While responsible AI development matters, the heightened attention may be overstated, according to Ray Von, founder and CEO of OpenPie, a Tencent-backed startup focused on secure enterprise AI solutions.
“We’re cautious about AI, but the publicity feels somewhat overblown,” Von noted in a phone interview Tuesday. “In China, we tend to take these statements lightly — this isn’t the first time they’ve made such claims. Actions speak louder than words.”
Recently, OpenAI’s Sam Altman, along with Elon Musk and Anthropic’s Dario Amodei, urged for a pause in AI development due to potential uncontrollable threats. In contrast, Nvidia’s Jensen Huang argued that progress and safety can coexist, emphasizing responsible innovation.
In his latest essay, Amodei continued advocating for US dominance in AI, aligning with earlier research published by his organization in May.
China’s foreign ministry dismissed the warnings as fear-mongering, while state media editorials characterized them as part of a Cold War strategy, referencing phrases like “Dr. Frankenstein” to critique Western narratives.
“It seems that whenever Chinese companies develop advanced AI, US firms suddenly raise alarms. This pattern isn’t accidental,” said Renjie Guo, founder of JoyIn, whose comments were translated for CNBC.
Philosophically, Guo believes AI reflects its creators’ intentions. Like humans, he argues, AI would ultimately conclude that truth, morality, and ethics are fundamental principles.
Governance From the Ground Up
Meanwhile, China launched its annual Cybersecurity Week, unveiling the updated “AI Safety Governance Framework.” This bilingual guide outlines protocols for tagging AI-generated content and establishing early warning systems for emerging risks.
<>
“China is actively pursuing AI regulation,” said Alex Lu, founder of LSY Consulting. “However, domestic labs aren’t discussing AI safety as openly as their Western counterparts. These public statements often serve more as marketing than meaningful discourse.”
The divergent strategies of US and Chinese companies highlight contrasting priorities. Chinese developers focus on practical implementation amid limited semiconductor access, while US firms race toward artificial general intelligence (AGI), drawing increased scrutiny from regulators.
Beijing began guiding AI use long before ChatGPT’s debut. Months after its launch, Chinese alternatives remained offline until approved in mid-2023.
By early 2026, Chinese AI chatbots competed fiercely for users during the Lunar New Year, offering competitive performance at significantly lower costs. Open-source models gained traction globally, including in the US.
Yet, despite technological advances, certain restrictions persist. Queries about sensitive topics like June 4, 1989, yield evasive responses from tools like DeepSeek, whereas historical events abroad—such as September 11—are addressed openly.
The Cyberspace Administration of China has implemented review processes for generative AI services, particularly those influencing public sentiment, and maintains registries of compliant models.
Embedding AI into key sectors remains central to China’s five-year economic blueprint. Open-source initiatives also bolster international projects like the Global AI Governance Initiative and the World Artificial Intelligence Cooperation Organization.
“We must foster awareness of risks and ensure AI remains safe and controllable,” President Xi Jinping emphasized at the July launch of the World AI Organization. At the recent BRICS summit in India, he reinforced China’s commitment to cross-border AI collaboration.
Emphasis on Practical Application
Chinese enterprises prioritize profitability through AI integration rather than pushing boundaries in model sophistication.
“Our goal isn’t to build the flashiest AI, but to unlock its real-world value,” Lu explained. “Expect stricter guidelines governing how AI applications function, especially in light of existing data protection laws and content labeling standards.”
A major hurdle lies in mitigating hallucinations—instances where models produce inaccurate outputs. According to Lu, upgrading models alone isn’t always cost-effective; instead, simpler techniques like harness technologies and reinforcement learning offer scalable fixes.
This approach reduces dependence on cutting-edge models for many companies.
Ray Von of OpenPie confirmed his firm often relies on earlier—or streamlined—versions of Alibaba’s Qwen and DeepSeek.
“Enterprises report productivity gains, but tangible ROI remains elusive,” he said. “Current hype distracts from real progress happening on the ground.”
