Chinese apps stop access to ChatGPT

Several Chinese apps have removed access to ChatGPT, the artificial intelligence chatbot developed by OpenAI. Despite not being officially available in China, some social media platforms had previously allowed access to the chatbot without a VPN or foreign mobile number. However, apps including ChatGPTRobot, AIGC Chat Robot, ChatgptAiAi, and Chat AI Conversation, have since suspended their programs. Shenlan BL gave no clear explanation for the shutdown. State-run media has also criticised ChatGPT, suggesting the chatbot could be used by US authorities to “spread disinformation and manipulate public opinion.”

It remains unclear what prompted the closures. Nevertheless, several Chinese tech companies saw their shares fall in response to the news. These include Beijing Haitian Ruisheng Science Technology, Hanwang Technology and Beijing Deep Glint Technology. Global concerns have also been raised about the unknown long-term consequences of the rise of ChatGPT, including its impact on education and students’ ability to cheat on assignments. The success of the chatbot has also spurred a global AI race, with Microsoft unveiling its AI-powered Bing chatbot last week and Google announcing the forthcoming rollout of Bard.

China’s government has previously been accused of digital protectionism for restricting access to major Western websites and apps. In the absence of foreign competition within the domestic market, Chinese tech companies have grown into major international players, many of which are now focusing on AI. For example, Alibaba has announced it is testing its own ChatGPT-style tool, while a team at China’s Fudan University developed their own version called MOSS, causing the platform to crash this week due to too many users. Furthermore, Baidu has announced its AI chatbot ERNIE Bot will be launched in March and used across various platforms.

The development of ChatGPT has brought concerns over privacy and ethics. In October 2020, a chatbot named GPT-3 was released, capable of generating convincing, realistic text. However, this was subsequently found to be able to generate harmful and misleading messages, including some containing hate speech. Despite these issues, OpenAI made GPT-3 available via an API to selected partners to build chatbots and other language models. The concerns over ChatGPT may suggest a wider debate on the regulation of AI-powered chatbots and similar technologies is required in the near future.