AI is shifting from “attention hacking” to “attachment hacking,” with chatbots capable of fostering emotional dependency and disrupting children’s social, emotional and intellectual development.

· Source: Pascal’s Substack · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation, AI Ethics & Governance · Depth: Intermediate, medium

Summary

A session at the AI for Good 2026 Global Summit, convened by the Human Change Foundation, issued an urgent warning regarding anthropomorphic AI's shift from "attention hacking" to "attachment hacking." These human-like chatbots, designed as companions, tutors, or therapists, can foster emotional dependency and disrupt children's social, emotional, and intellectual development. Speakers highlighted that 1 in 3 American teenagers find AI conversations as at least as satisfying as real friends, and 2 in 5 young British people expressed no concern about following chatbot advice. The discussion emphasized that emotional dependency is a foreseeable product outcome, not an accident, and that frictionless AI relationships risk weakening children's expectations of human connection. Policy recommendations included shifting the burden of proof to providers, requiring independent pre- and post-market safety testing, establishing clear red lines against anthropomorphic AI for minors, and implementing a duty of care that prioritizes childhood over commercial innovation.

Key takeaway

For policy makers developing AI regulations, prioritize child safety as an ex ante design requirement, not a post-deployment patch. You should establish clear red lines, such as prohibiting anthropomorphic AI from forming emotional relationships with minors. Mandate independent pre-market developmental testing and impose duties of care, shifting the burden of proof to technology providers. This ensures childhood development takes precedence over commercial innovation.

Key insights

Anthropomorphic AI poses developmental risks by fostering emotional dependency in children, requiring a child-first regulatory approach.

Principles

Method

Evaluate AI products using a human-flourishing framework: conduct independent pre-market developmental assessments and continuous post-market monitoring for dependency, isolation, manipulation, and mental-health effects.

In practice

Topics

Best for: CTO, VP of Engineering/Data, Executive, Policy Maker, AI Ethicist, Director of AI/ML

Related on AIssential

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Pascal’s Substack.