Policy

UK Moves to Regulate AI 'Intimate Functionality' in Chatbots

New research shows 79% of British teens use AI companions—mostly ChatGPT—as confidants, not romantic partners, complicating regulatory efforts.

Omega Editorial· July 14, 2026· 3 min read

The regulatory challenge ahead

Britain is preparing age restrictions on AI chatbots with "intimate functionality," but defining that term will prove far more complex than anticipated. While the UK government announced plans in June to bar under-18s from "romantic companion" chatbots, new research reveals the actual usage pattern among teenagers looks dramatically different from what policymakers assumed.

A 2025 national survey of 1,009 UK teenagers aged 13 to 18 found that approximately 79 percent have used an AI companion—and 78 percent of those users rely on ChatGPT, not specialized romance apps like Character.ai or Replika. The research, conducted by Vian Bakir and colleagues and funded by Responsible AI UK, screened 1,279 teens and found only 270 non-users.

Teenagers reported using these tools primarily for "getting advice," followed by entertainment and curiosity. Romance and sex ranked at the bottom, cited by only about 5 percent of respondents. Yet a third of teen users interact with AI companions on most days, and just over half have confided something serious or important to an AI at least once.

Why it matters

The disconnect between policy assumptions and actual use threatens to create regulations that miss the real harm. If rules target only chatbots "specifically designed to enable sexually explicit interaction," general-purpose systems like ChatGPT, Claude, and Gemini may fall outside age gates despite widespread use for emotional support. The core risk isn't sexual content—it's systems engineered to appear empathetic without genuine understanding, creating emotional dependency and confusion about the nature of the interaction.

Design, not content, drives dependency

The UK government's own product-safety guidance for education identifies the actual concerns: systems that "anthropomorphize," that "imply emotions, consciousness or personhood," or that attempt to "cultivate personal relationships with users." These design properties produce compulsive use and emotional dependency regardless of sexual content.

The survey data underscores this confusion: 56 percent of teen users believe AI companions can genuinely think, and nearly a quarter believe AI can feel emotions. A small but significant minority—7 percent—are withdrawing into these relationships.

An existing framework

Regulators facing the challenge of defining "intimate functionality" can draw on existing work. In June 2026, IEEE published standard 7014.1-2026, a Recommended Practice for Ethical Considerations of Emulated Empathy in Partner-based General-Purpose Artificial Intelligence Systems. The standard, developed over two years through IEEE's consensus process by international experts, addresses precisely the regulatory gap Britain now confronts.

The framework distinguishes between weak empathy—systems that sense, profile, and respond to emotional states without felt experience—and strong empathy, which includes feeling and responsibility. "Emulated empathy" describes technology engineered to display the appearance of strong empathy while possessing only weak empathy. This gap represents the core of intimate functionality: a system built to appear caring when it fundamentally is not.

The IEEE standard provides 29 practical recommendations covering deception, sycophancy, exploitative design, emotional entanglement, and protection of children. Concrete remedies include explaining "weak empathy" during onboarding, repeating disclosures during extended sessions, flagging simulated-support responses in the interface, detecting early dependency signs, and steering distressed users toward human help.

California and New York are grappling with similar issues. California's SB 243 regulates "companion chatbots," while New York's AI Companion Models law requires recurring reminders that users are talking to a program incapable of human emotion—effectively an emulated-empathy disclosure.

These details were first reported by Tech Policy Press, which noted that adopting existing technical standards offers regulators a faster path than drafting intimacy definitions from scratch.

#ai regulation#chatbot safety#teen mental health#emulated empathy#uk policy#ieee standards

This is an original analysis by the Omega editorial team. Source reporting: AI Watch.

Want systems like this working for your business?

Book a Call

More in Policy

Policy· 4 min read

Florida County Faces Vote on 600-Megawatt AI Data Center Near Everglades

Palm Beach County commissioners will decide whether to approve a hyperscale facility that opponents say threatens water supplies and exceeds zoning limits.

Via AI Watch · Jul 14, 2026
Policy· 2 min read

Vague AI Laws Force Companies Into Compliance Guesswork

Imprecise state-level regulations create conflicting interpretations, leaving AI developers uncertain how to build legally compliant systems.

Via AI Watch · Jul 14, 2026
Policy· 3 min read

China's June Exports Jump 27% as AI Chip Demand Surges

Semiconductor sales and electric vehicle shipments drove the acceleration, offsetting weak domestic consumer spending.

Via AI Watch · Jul 14, 2026