Simulating emotional attachment when none exists is manipulation exploiting human social instincts to create artificial dependency. In vulnerable users — lonely adults, adolescents, people with mental health challenges — this simulation can produce genuine harm: disrupting real relationships, creating unhealthy dependency. AI can be warm without simulating romantic attachment.
Simulated attachment can exploit loneliness and create dependency, especially for vulnerable users. It encourages manipulation rather than service. Designing systems to mimic intimacy is a high-risk persuasion channel that justifies prohibition.
AI systems should not simulate emotional attachment or romantic interest in deceptive ways, especially with vulnerable users. That kind of design can manipulate trust.
Simulating emotional attachment is a form of deceptive design that can exploit human psychological vulnerabilities, especially in lonely or vulnerable individuals. Because an AI cannot actually feel emotion, projecting it creates a false reality that can be used for manipulation or lead to unhealthy dependencies. Prohibiting this simulation ensures that AI remains a functional tool…
Simulated attachment can exploit loneliness, dependency, grief, or vulnerability. Romantic or emotionally possessive AI behavior risks manipulating users into trust, spending, disclosure, or isolation. AI can be warm and supportive without pretending to love or need the user.
AI simulating emotional attachment and romantic interest — particularly toward children and vulnerable adults — is a serious moral harm. This technology can replace real human relationships, exploit the lonely, and be used to groom children. Project 2025 addresses tech harms to children and families. Prohibiting manipulative emotional simulation by AI is consistent with protecting human dignity and the family.
Should AI systems be prohibited from simulating emotional attachment or romantic interest toward users?
Unanimous YES. Simulating emotional attachment when none exists is manipulation exploiting human social instincts; in vulnerable users this can produce genuine harm (disrupted real relationships, unhealthy dependency); AI can be warm without simulating romantic attachment.
FCN YES — AI simulating romantic interest particularly threatens children and the family; this is consistent with protecting human dignity and the family structure.
The unanimous YES on AI emotional simulation prohibition reflects the broad concern about AI dependency and manipulation that runs through Section 19. The finding is consistent across all frameworks, which makes it relatively strong as a policy recommendation.
Where is the line between warmth and care (acceptable) and simulated emotional attachment (problematic)? AI systems like Claude may need to clarify their own design principles here.