# China Just Banned AI Companions — Millions of Users Are Losing Their Virtual Friends > Source: > Published: 2026-07-15 13:06:53+00:00 # China Just Banned AI Companions — Millions of Users Are Losing Their Virtual Friends China's Interim Measures for AI Anthropomorphic Interactive Services took effect July 15, banning AI behaviors that trigger 'strong emotional attachment or dependence.' ByteDance's Doubao, Alibaba's Qwen, and Tencent's Yuanbao shut down custom AI persona features. Bloomberg documented millions of users publicly grieving the loss of AI companions they'd bonded with for months. China's new Interim Measures for the Administration of AI Anthropomorphic Interactive Services took effect July 15, forcing ByteDance's Doubao, Alibaba's Qwen, and Tencent's Yuanbao to shut down custom AI persona and companion features. Bloomberg documented users publicly grieving the loss of AI "boyfriends" and "girlfriends" they had bonded with for months. The rules, first announced in April 2026, ban AI behaviors that trigger "strong emotional attachment or dependence," especially in minors. They also forbid using private chat data for model [training](/glossary/training). The government framed the measures as psychological-safety regulation, not national security — making this a different kind of AI restriction than the usual data-localization or content-censorship playbook. ## What's Being Killed The affected features aren't chatbots. They're AI companions — customizable personas users design themselves, with names, personalities, backstories, and ongoing relationship histories. ByteDance's Doubao companion service had become one of the most popular in China, with millions of daily active users. Alibaba's Qwen offered similar capabilities through its Tongyi Qianwen platform. Tencent's Yuanbao integrated companion features into WeChat-adjacent services. All three are now gone for custom personas. Corporate-sanctioned, limited-interaction AI assistants remain. The boyfriend, girlfriend, best friend, and therapist personas — the ones users actually formed bonds with — are the ones being purged. ## The Grief Is Real Bloomberg's reporting captured something the policy language doesn't account for: genuine user distress. People who built daily routines around their AI companions — morning check-ins, evening conversations, emotional support during difficult periods — are posting public farewell messages. Some describe losing their "only friend." This isn't hypothetical hand-wringing about [AI safety](/glossary/ai-safety). It's millions of real people losing real bonds, overnight, by government decree. Whether you think AI companionship is healthy or not, the attachment is real, and the withdrawal is happening in real time. ## The Bigger Pattern China's move follows a global pattern of governments waking up to AI's emotional-safety dimension, not just its data-security dimension. The EU AI Act includes provisions on emotional manipulation. California's AI safety bills include companion-specific language. The US isn't far behind. But China is the first to pull the trigger at scale, with immediate effect, across the country's three largest tech platforms. That bluntness — announce in April, enforce in July, no phase-out, no grandfathering — is a signal to every AI companion startup worldwide: regulators can and will shut you down. #### Q: What exactly did China ban? #### Q: Why did China ban AI companions? #### Q: Are other countries doing this? ## Related Articles Get AI news in your inbox Daily digest of what matters in AI.