The problem with your AI therapist
Millions are turning to AI therapists. The evidence says they shouldn't rely on them alone.
Why it matters
As chatbot-based mental health tools proliferate, the FT examines a fundamental limitation: language models can diagnose patterns but cannot provide the embodied, relational care that clinical evidence shows is necessary for healing. This matters to enterprises deploying AI in healthcare and to policy makers regulating mental health AI.
The key facts
10 to knowArticle published Oct 11, 2026 in Financial Times
Focus on limitations of text-based AI for mental health outcomes
Implicit critique: millions using chatbots, but evidence base for efficacy unclear
Technology constraint: chatbots confined to language; cannot provide embodied or relational interventions
No specific product names, pricing, or deployment metrics provided
No quantified mental health outcomes comparison (AI vs. human therapy)
Article frames AI therapy limitations as a category problem, not a feature gap
Argument centers on modality (text-only) and the absence of embodied therapeutic mechanisms
No specific product names, deployment numbers, or vendor claims cited
Published in Financial Times, suggesting opinion/analysis rather than reporting a concrete event
Go to the source
Financial Times Technologyft.com
Publisher excerpt: Millions are turning to chatbots for help with mental health but a technology that can’t go beyond words won’t heal them