Some of China’s most popular artificial intelligence (AI) chatbot apps, made by leading developers including Zhipu AI and Moonshot AI, had been collecting excessive user information, authorities found.

Zhipu’s Qingyan, also known as ChatGLM, was found to have collected information beyond what users authorised, while Moonshot’s Kimi had accessed data irrelevant to its functions, according to a list published on the WeChat account of the National Cyber Security Information Centre on Tuesday.

Neither Zhipu nor Moonshot, start-ups that are both based in Beijing, immediately responded to requests for comment on Wednesday.

ChatGLM and Kimi ranked among the 10 most popular AI applications in the country, boasting a combined total of nearly 35 million monthly active users as of April, according to Aicpb.com, which tracks the popularity of AI apps and services