Smart City Governance and Public Health in Shenzhen

In Shenzhen, where nearly 450 AI medical products have already been deployed across city healthcare institutions and AI large models now assist in diagnosing cases across 30 top-tier hospitals, the integration of artificial intelligence into daily health decisions has become unavoidable. AI services, with the most prominent example being large language model (LLM) chatbots such as Doubao, are becoming the go-to consult for residents in face of immediate health anxieties. This convenience, once reserved for clinic visits, outlines  chatbot’s promising potential as an accessible public health service, yet beneath this technological optimism lies an unresolved question: in critical health situations, who would be responsible to regulate the chatbot’s hallucinations and misinformation? As AI becomes the inevitable first point of contact for health information in Shenzhen, the regulation of AI-driven health chatbots is not an unreachable speculation, but rather an urgent public health imperative that must be resolved. 

On this account, effective regulation on health-related LLM chatbots must be enacted to prevent unseen emergencies and dangers that misinformed health advice impose on residents in Shenzhen.

A recent scholarly evaluation of AI chatbots found that 88% of queried customized chatbots propagated malicious false health information when prompted (Modi et al. 4), a statistic that should alarm any policymaker overseeing a rapidly digitizing city like Shenzhen. In the core, unregulated models lack the clinical accountability frameworks that govern human medical professionals. Shenzhen, despite establishing China’s first dedicated AI and Robotics Administration in Longgang District in March 2025 (China.org.cn), has yet to implement comparably specific safeguards for consumer-facing health AI.

The vulnerability of certain populations compounds this risk significantly. Older adults and individuals with limited digital literacy are particularly susceptible to AI hallucinations, which are confidently delivered yet factually incorrect outputs — and can unknowingly propagate this generated misinformation within their social networks (Zhu et al. 7). This is especially salient in the Greater Bay Area, where an aging demographic intersects with high rates of smartphone and messaging app adoption. The dangers to the phenomenon is multifaceted, as this demographic section also experiences higher proportions of health issues and health related anxieties.

Such risks are highly recommended to be managed with the support of the highly technological identity of Shenzhen itself, as it gradually cements itself as the center of technological businesses and the top-end AI research hub in China. Potential measures included the increased customization of chatbots for health related purposes, such as CUHK/Shenzhen’s Huatuo GPT AI medical assistant, already deployed in 12 public hospitals and 183 community health centers serving over 600,000 residents (Zhang et al.). Furthermore, the city of Shenzhen can further integrate this concern with its highly developed AI research framework, collaborating and regulating relevant service providers to provide accurate and helpful health-related information.

AI health chatbots are neither the cure nor the disease, but a potent tool whose impact depends entirely on the governance structures surrounding them. Shenzhen and the broader Greater Bay Area, as the proving ground for China’s AI ambitions, have both the responsibility and the resources to model what regulated, human-centered AI health communication can look like (China.org.cn). Authorities should mandate transparency disclosures, enforce clinical accuracy standards for health-adjacent LLMs, and invest in AI literacy programs specifically targeting older adults and digitally underserved communities, turning Shenzhen’s AI pioneer status into a genuine public health advantage.

By Timmy Zhang

Bibliography

China.org.cn. “How China’s Shenzhen Is Transforming into an AI-Powered City.” China.org.cn, 23 Mar. 2026, www.china.org.cn/china/Off_the_Wire/2026-03/23/content_118397436.shtml.

Modi, R., et al. "Study Finds AI Chatbots Are Vulnerable to Spreading Malicious False Health Information." The ASCO Post, 5 Jun. 2025, https://ascopost.com/news/june-2025/study-finds-ai-chatbots-are-vulnerable-to-spreading-malicious-false-health-information/. Accessed 26 Apr. 2026.

Zhang, Hongbo, et al. “HuatuoGPT, towards Taming Language Model to Be a Doctor.” ArXiv.org, 2023, arxiv.org/abs/2305.15075.

Zhu, K., et al. "Empowering Alzheimer’s caregivers with conversational AI." PubMed Central, vol. 1, no. 1, 2024, https://pmc.ncbi.nlm.nih.gov/articles/PMC13055072/. Accessed 26 Apr. 2026.