According to a recent survey, half of all U.S. adults now use AI chatbots, up from a third in 2024, and one in ten reported using chatbots for emotional support.
In another study, which included data from Germany, China, South Africa, and the U.S., at least one-third of users reported attachment-related behaviors when interacting with general-purpose AI chatbots, including feeling less lonely, freedom from judgment, and a sense of privacy.
Dedicated platforms for AI companionship, such as Replika and CharacterAI, have also amassed tens of millions of users as human-AI emotional bonding and intimacy become less stigmatized, with cases of human-AI romances, including marriages, on the rise.
However, emotional attachment between humans and AI is not without risks.
It can lead to dependence, social and moral deskilling, social isolation, and the exacerbation of existing mental health issues.
Especially among children and other vulnerable groups, and particularly when human-AI interactions are continuous and emotionally intense, they could lead to severe harm, with multiple cases of AI-chatbot-related suicides documented.
Aiming for user loyalty, increased usage time, and long-term profitability, many companies have been optimizing the anthropomorphic features of their AI systems.
Companies have also been called out for exaggerating human-like features, exploiting human affection, and failing to oversee the risks and mitigate the harm from human-AI interactions.
With that in mind, and to help companies, AI professionals, policymakers, and researchers understand the broader implications of anthropomorphic AI systems and how to address them comprehensively, today I’m launching my AI Policy Masterclass on “Governing Human-Like AI.”
In this Masterclass, I analyze and contextualize China’s recently enacted law, the world’s strictest and most comprehensive legal framework for human-like AI, and explain how it can inspire AI governance efforts in this area.
This Masterclass and the accompanying reading list will be available to paid subscribers until September 4. Click the image above to watch it (59 minutes).
👉 You can find the recommended reading list for this Masterclass here.
