China Tightens AI Rules: Companion Chatbots, Labeling, and a Comprehensive Law
In 2026 China advanced a dedicated rule for AI companion chatbots (effective 15 July), kept its content-labeling regime in force, and signaled a comprehensive national AI law is coming.
China spent 2026 extending one of the world’s most detailed AI rulebooks. A dedicated rule for AI companion and emotional-interaction chatbots takes effect on 15 July 2026, the mandatory content-labeling regime that went live in September 2025 continues to be enforced, and the State Council signaled that a single comprehensive national AI law is on the way. For anyone serving Chinese users, the obligations are concrete and expanding.
What happened
The Cyberspace Administration of China (CAC), with four other bodies, promulgated final Interim Measures for the Management of Anthropomorphic Interactive Services of AI (Decree No. 21) on 10 April 2026, effective 15 July 2026. Providers of AI companion or emotional-interaction services must clearly tell users they are talking to AI, and must trigger reminders about over-dependence at first use, on re-login, and when addiction signals appear. It is the first national rule aimed specifically at AI companion chatbots.
This sits on top of an existing base. The Measures for the Identification of AI-Generated and Synthetic Content, effective since 1 September 2025 with the mandatory standard GB 45438-2025, require both visible labels and metadata on AI-generated content, and enforcement has already begun. In May 2026 the State Council’s annual legislative plan called to “accelerate comprehensive legislation” on AI, the clearest signal yet of a single overarching law. China also hosted the 2026 World AI Conference in Shanghai on 17 to 20 July, continuing to promote its Global AI Governance Action Plan.
Why it matters for builders
China’s approach is sectoral and fast-moving: rather than one big act, it issues targeted, binding measures, and it issues them quickly. If your product touches Chinese users, three obligations are live or imminent: label AI-generated content (in force), disclose and add dependency safeguards on companion or emotional chatbots (from 15 July), and expect a comprehensive law to consolidate these in future.
The contrast with the EU (one broad regulation) and Korea and Japan (Korea comprehensive, Japan light-touch) is the point: there is no single global standard, and China’s is among the most prescriptive on synthetic media and companion AI. Build to the specific rules of each market you serve.
Sources
- CAC, final Interim Measures for Anthropomorphic Interactive Services of AI (10 April 2026, effective 15 July 2026): https://www.cac.gov.cn/2026-04/10/c_1777558395078289.htm
- CAC, Measures for the Identification of AI-Generated and Synthetic Content (effective 1 September 2025): https://www.cac.gov.cn/2025-03/14/c_1743654684782215.htm
- State Council 2026 Legislative Work Plan (11 May 2026): https://www.gov.cn/yaowen/liebiao/202605/content_7068353.htm
- State Council (English), 2026 World AI Conference, Shanghai (13 July 2026): https://english.www.gov.cn/news/202607/13/content_WS6a5494bcc6d00ca5f9a0c27c.html
Further reading
- Beyond the EU: South Korea and Japan’s AI laws : the other major Asian regimes.
- EU AI Act: what takes effect on 2 August 2026 : the comprehensive-regulation model.
- The US moves to preempt state AI laws : a third, litigation-driven approach.