Love me like AI does
It carries the pathos of a 21st-century tragedy, and an indictment even of modern dating. The machine, it appears, is acing the game of love. Given the overwhelming response to AI companionship, China has introduced regulations forbidding chatbots from cultivating emotional dependence, banning virtual romances with minors, and requiring companies to issue alerts when users display signs of emotional distress. These measures are not merely about taming an emerging technology. Against the backdrop of an ageing population — China’s population declined for the fourth consecutive year in 2025, while the birth rate fell to a historic low — they also reflect an attempt to nudge people back towards relationships rooted in the real world. Yet that is easier said than done. AI companions are engineered to be infinitely patient and emotionally fluent; unlike human beings, they possess the miraculous ability to say the right things at the the right moment. Not for them the awkwardness of forgotten birthdays, bitter misunderstandings, and tetchy exchanges that begin with “we need to talk”. The algorithm offers companionship stripped of inconvenience or the uncomfortable possibility of rejection. In societies where loneliness is increasingly pervasive, such frictionless intimacy becomes formidable competition for the imperfect business of loving another person. China is not alone in grappling with this reality. States such as New York and California in the US have introduced measures requiring chatbots to periodically disclose that they are not human. Such safeguards acknowledge a truth: As artificial companionship grows more persuasive, the real contest may not be between humans and machines, but relationships that demand vulnerability and those that promise affection without risk.
- 1China's ability to swiftly mandate emotional-safety disclosures for AI chatbots reflects its unitary, single-party governance structure, where the Cyberspace Administration of China can issue binding rules with limited institutional friction. India, by contrast, would need to route comparable AI safety rules through its federal structure and existing frameworks like the Information Technology Act, 2000 and its intermediary guidelines. This contrast illustrates how a country's constitutional design, whether centralized or federal, shapes the speed and mechanism of emerging technology regulation.
- 2The editorial highlights a global pattern of divergent AI governance: China's top-down ban on emotional dependence and minor-targeted virtual romance contrasts with the narrower, disclosure-based approach of US states like California and New York. This divergence mirrors the broader China-US contest over technology standard-setting, seen also in debates over data localization and semiconductor export controls. Because AI companion apps operate across borders, such regulatory fragmentation raises questions about which jurisdiction's rules ultimately govern global user protection.
- 3China's chatbot rules are enforced through its Cyberspace Administration, drawing on the framework of the Personal Information Protection Law of 2021 and prior generative-AI regulations issued in 2023. In the United States, California's chatbot disclosure law and similar New York State legislation require AI companions to identify themselves as non-human, a narrower consumer-protection approach compared to China's broader emotional-dependence ban. India's own Digital Personal Data Protection Act, 2023, and the proposed Digital India Act are the closest domestic analogues, though neither yet specifically addresses AI companionship risks.
- 4China's population has now declined for four consecutive years as of 2025, with its birth rate falling to a historic low, a demographic reality the editorial links directly to the rise of AI companionship as a substitute for human relationships. Rapidly aging societies, including Japan and South Korea, have faced similar debates over technology filling emotional and caregiving gaps left by shrinking family structures. The rapid growth of large language models capable of sustaining long, emotionally responsive conversations has made AI companion apps far more convincing than earlier chatbot generations.