A Law Aimed at AI Girlfriends Just Wiped 8 Million Agents From China’s Biggest AI App
Last Updated on July 16, 2026 by Editorial Team
Author(s): Chew Loong Nian – AI ENGINEER
Originally published on Towards AI.
A Law Aimed at AI Girlfriends Just Wiped 8 Million Agents From China's Biggest AI App
At midnight Beijing time today, ByteDance’s Doubao — roughly 350 million monthly active users, the second most-used AI app on Earth behind ChatGPT — switched off its user-built agent feature. More than 8 million custom agents, per ByteDance’s last public disclosure, went read-only. Users have until October 15 to screenshot or export their conversations. After that, the data is gone for good.

After the lead, the article ties the shutdown to China’s new interim rules on anthropomorphic AI interaction services, showing how multiple major apps (Tencent’s Yuanbao, Alibaba’s Qwen, and now Doubao) removed user-built “companion” agents even when the law says productivity agents are exempt. It explains the core problem: regulators define the boundary behaviorally (who the system induces sustained emotional interaction), which makes auditing millions of user-created agents intractable, so platforms choose “compliance by amputation” (deleting the whole user-agent feature) rather than classifying and remediating edge cases. The piece then maps what the law requires—ongoing “I am an AI” disclosure, usage interruption reminders, an immediate exit option, crisis/persona break handling for acute distress, and restrictions for minors—contrasting these with how persistent-memory companions are designed to deepen attachment. It broadens the scope beyond China by noting similar legal pressures in California and the EU AI Act, and concludes with practical engineering takeaways (wrapping the agent loop with middleware for exit, crisis routing, disclosure timing, and exportable memory) plus a “verdict” predicting that memory will become tiered/separable, compliance will spread via deletion of fuzzy user-built persona features, and exemptions will guide product roadmaps toward agents that “do things” over those that “feel things.”
Read the full blog for free on Medium.
Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.
Published via Towards AI
Towards AI Academy
We Build Enterprise-Grade AI. We'll Teach You to Master It Too.
15 engineers. 100,000+ students. Towards AI Academy teaches what actually survives production.
Start free — no commitment:
→ 6-Day Agentic AI Engineering Email Guide — one practical lesson per day
→ Agents Architecture Cheatsheet — 3 years of architecture decisions in 6 pages
Our courses:
→ AI Engineering Certification — 90+ lessons from project selection to deployed product. The most comprehensive practical LLM course out there.
→ Agent Engineering Course — Hands on with production agent architectures, memory, routing, and eval frameworks — built from real enterprise engagements.
→ AI for Work — Understand, evaluate, and apply AI for complex work tasks.
Note: Article content contains the views of the contributing authors and not Towards AI.