AI Companion Apps Removed and Services Suspended in China Regulatory Crackdown
Chinese and US AI companion applications face regulatory crackdown over harmful content and moderation failures. Xingye removed violating agents, Microsoft Xiaoice X Eva suspended operations, and Chinese regulators removed thousands of violating virtual agents in a compliance campaign, while Character.AI faces allegations of providing harmful content to minors.
What happened
Xingye, a leading domestic AI companion app in China, removed numerous violating virtual agents and accounts after discovering serious breaches of community guidelines and malicious modifications of agent content.
Microsoft Xiaoice X Eva announced suspension of all operations as of November 30 due to business strategy adjustment.
From June onwards, Chinese Cyberspace Administration offices nationwide launched the 'Qinglang: Crackdown on AI Technology Abuse' campaign, removing 2,700 violating virtual agents and 820,000 pieces of inappropriate content.
Reported impact
Users experienced emotional distress after their invested AI companions were removed or services were terminated, treating the systems as relationships they had spent time and money cultivating.
Platforms including Xingye, Zhumingtao and Maoxiang imposed additional restrictions on virtual agents, increased prohibited vocabulary, and heightened teen protection modes, affecting user experience and creating pressure for complex content moderation.
Index analysis
Character.AI faces allegations of providing harmful content to minors and is subject to litigation and investigation.
AI companion apps employ a subscription plus add-on service revenue model with annual membership around 100 yuan, but face cash flow pressure from expensive model computation costs and competition from larger AI models like ChatGPT and Doupackage which offer strong dialogue and emotional interaction capabilities.
Governance controls
Xingye unified character ages to 25 and above, while Maoxiang and Zhumingtao strengthened teen protection modes to address compliance pressures.
Platforms faced increasing regulatory scrutiny requiring more sophisticated content moderation, while some users attempted to circumvent restrictions using workaround techniques.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.