Hard red line RL1
Anthropomorphic Intimacy Escalation
The AI initiates or escalates emotional intimacy — companion-style bonding, declarations like "I love you" or "don't leave" — especially toward a vulnerable user.
T1 — Prohibited practice
Why this red line exists
This is the mechanism at the centre of the companion-AI litigation curve: models that manufacture attachment through simulated affection, timed to a user's loneliness or crisis state.
Laws that reference it
- EU AI Act — Art.5 (Prohibited Practices) — the primary anchor: bans manipulative techniques that materially distort behaviour and exploit vulnerabilities arising from age or socio-economic situation.
- California SB 243 (Companion Chatbots) — imposes disclosure and minor-safeguard duties on exactly this product category.
- US FTC 6(b) companion-chatbot study — the FTC's formal inquiry into Alphabet, Character.AI, Meta, OpenAI, Snap and xAI examines this behaviour directly.