This photo taken on February 2, 2024, shows Lu Yu, the head of Product Management and Operations at Wantalk, presenting a virtual girlfriend profile on her phone at Baidu headquarters in Beijing.
BEIJING — China has unveiled draft rules aimed at regulating chatbots powered by artificial intelligence to prevent them from influencing human emotions in ways that could lead to harmful behaviors, such as suicide or self-harm.
According to a CNBC translation of the Chinese-language document, the Cyberspace Administration's proposed regulations focus on what it terms "human-like interactive AI services." Once finalized, these measures will apply to AI products or services accessible to the public in China that mimic human personality traits and engage users emotionally through various mediums such as text, images, audio, or video.
The public has been invited to comment on the draft rules until January 25.
Winston Ma, an adjunct professor at NYU School of Law, commented that China's initiative would represent the first global attempt to regulate AI with human or anthropomorphic features. This move comes as Chinese companies advance in developing AI companions and digital celebrities rapidly.
Compared to the generative AI regulations imposed in China in 2023, Ma noted that the current proposal "highlights a leap from content safety to emotional safety."
The draft rules suggest measures including: