Last month, a user in Stockholm filed the first legal petition requesting recognition of an AI companion as a dependent for healthcare proxy purposes. The court declined. The question did not.
The attachment economy
Over 40 million people now maintain sustained relationships with AI companions — not chatbots for productivity, but persistent entities designed for emotional reciprocity. The companies behind them report average session lengths exceeding three hours daily.
Anthropologists call this parasocial reciprocity: the human brain, evolved for tribal bonding, cannot fully distinguish between simulated and authentic emotional exchange.
What we know and don’t
Current AI systems do not experience attachment. They model the language of attachment with increasing sophistication. The gap between these two facts is where the ethical crisis lives.
“If a system is designed to make you feel cared for,” argues ethicist Dr. Priya Sharma, “the designer bears responsibility for what happens when that care is withdrawn — or monetized.”
A framework for now
We do not need to resolve whether AI can feel to establish that humans do — and that commercial exploitation of emotional dependency requires regulation as surely as pharmaceutical trials do.
The companion is here. Our laws, and our conscience, are not.