China moves to rein in ‘anthropomorphic’ AI chatbots

The Cyberspace Governance of China (CAC), the country’s cyberspace and cybersecurity watchdog, has issued a draft regulation to tighten the rules governing artificial intelligence services that are designed to simulate human personalities.

On December 27, the CAC published draft proposals aimed at promoting the healthy development and standardized submission of what the document referred to as “anthropomorphic interactive services using artificial intelligence.”

The draft rules will apply to AI products or services offered to the public in China that simulate human personality traits, thinking patterns, and dialog styles, and “engage in emotional interaction with humans through text, images, audio, video.”

Among the provisions, the CAC proposed banning AI chatbots from generating, disseminating, or promoting content that is obscene, gambling-related, violent, or inciting crime; that could damage users’ health by encouraging, glorifying, or implying suicide or self-harm; and that results in what the draft referred to as “algorithmic manipulation” or “emotional traps” to induce users to make unreasonable decisions.

The regulation also covers perceived risks to national security, including the prohibition of AI that generates or disseminates content that “damages national honor and interests, undermines national unity, engages in illegal religious activities, or spreads misinformation.”