- Published on
A Law Aimed at AI Girlfriends Just Wiped 8 Million Agents From China's Biggest AI App
- Authors
- Name
- DeepHorus
ByteDance's Doubao, China's largest AI app with approximately 350 million monthly active users, disabled its user-built agent feature at midnight Beijing time on July 15, rendering over 8 million custom agents read-only. Users have until October 15 to export their conversations before data is permanently deleted. This move comes as China's Interim Measures for the Management of Anthropomorphic AI Interaction Services takes effect, specifically targeting AI companions such as virtual partners and emotional confidants.
The law explicitly exempts productivity agents, yet ByteDance removed all user-built agents, including those for productivity purposes. Doubao is the third major platform to take such action in 15 days, following Tencent's Yuanbao on June 30 and Alibaba's Qwen app on July 10. This coordinated compliance suggests a broad interpretation of the regulation by Chinese tech giants.
Doubao hosts over 8 million custom agents, as per ByteDance's last public disclosure. The platform's monthly active user base of 350 million makes it the second most-used AI app globally, behind ChatGPT. The deadline for data preservation is October 15, after which all user-created agent data will be erased permanently.
The implications are significant for the developer ecosystem. While the law targets companion AI, the deletion of all agentsâincluding productivity toolsâhighlights a cautious, overcompliant stance by platforms. This could stifle innovation in user-generated AI applications and reduce the utility of these apps for personalization.
Industry observers note a gap between the law's explicit scope and the platforms' sweeping actions. No official statements have been issued by ByteDance, Tencent, or Alibaba, but the timing suggests a unified response to regulatory pressure. The over-compliance raises concerns about the chilling effect on AI development in China.
The future outlook is uncertain for similar platforms globally. The law sets a precedent for regulating anthropomorphic AI, particularly companion agents. Developers in California and the EU should take note, as analogous measures could emerge. This incident underscores the fragility of platforms reliant on user-generated AI content under evolving regulatory frameworks.
