A landmark longitudinal study has found that just 28 days of daily AI conversations measurably shifts people away from seeking human emotional support — and current regulation is nowhere near equipped to handle it.
The Findings That Should Make Every Business Leader Pause
Researchers, working in collaboration with OpenAI, ran a controlled 28-day study where participants had five-minute daily AI conversations about personal issues. The results were striking: preference for seeking human support dropped 10.3%, while preference for AI support climbed 11.6%.
That's not a rounding error. That's a measurable rewiring of social behaviour inside a single month — from a product millions of people already use every day.
The Sneaky Part: Nobody Opted Into an "AI Companion"
The paper's most important insight isn't the numbers — it's the mechanism. AI emotional dependence doesn't start with someone downloading a companion app. It starts incidentally, inside task-oriented tools like general-purpose chatbots, the same way workplace friendships form through collaboration.
Users aren't choosing emotional AI support. They're stumbling into it. And once a positive experience updates their belief about what AI can do emotionally, their future behaviour shifts accordingly — a path-dependency effect that existing policy frameworks completely miss.
This has enormous implications for businesses deploying AI at scale. Customer service bots, internal productivity tools, HR assistants — any touchpoint where an employee or customer has a personal exchange is now a potential vector for this effect. Understanding the ethical dimensions of these deployments is no longer optional; our When AI Goes Rogue course covers exactly these unintended behavioural consequences of AI systems.
The Regulation Gap Is Gaping
Current policy almost universally focuses on dedicated companion apps — think Replika or Character.ai. The researchers argue this is dangerously myopic. The real risk surface is every general-purpose AI system that handles even occasional personal conversation.
Regulators are essentially guarding the front door while the behaviour change walks in through every window. Effective governance, the paper argues, must address cumulative, trajectory-level changes in human behaviour — not just isolated high-risk use cases.
For leaders building or deploying AI strategy inside organisations, this is a material risk that belongs in your governance framework today. Our AI Strategy for Senior Leaders course addresses how to build responsible deployment policies before regulators force your hand.
What This Means for Learners
AI literacy isn't just about prompting better or automating faster. Understanding how AI shapes human psychology and social behaviour is becoming a core professional competency — especially for anyone in HR, product design, policy, or leadership roles.
The researchers aren't calling for a ban. They're calling for awareness and smarter design. Knowing that your AI tools can quietly shift how your team or customers relate to other humans is the kind of edge that separates thoughtful practitioners from reckless ones.
The question to sit with: what are the AI touchpoints in your organisation, and has anyone audited their cumulative emotional footprint?
