Survey Reveals 61% of People Use AI for Mental Health Support, But Safety Concerns Persist
Key Takeaways
- ▸61% of respondents globally use AI for mental health support, with adoption highest in China, the Philippines, and Turkey
- ▸32% of AI users have felt uncomfortable with advice received, and 28% engaged in harmful behavior based on AI recommendations
- ▸AI fills a critical gap for people avoiding professional help due to cost, time constraints, and stigma—but cannot replace professional mental health care
Summary
A new AXA and Ipsos survey across 18 countries reveals that more than 6 in 10 people are turning to AI for psychological support, even as global mental health continues to deteriorate. The report found that 46% of respondents report struggling or languishing mentally, with two-thirds believing screens negatively impact their well-being. While AI appeals to users as a free, 24/7 alternative to professional mental health services—addressing barriers like cost and stigma—the data also uncovers significant risks.
Of those using AI for mental health advice, 42% say they almost always follow its recommendations. However, 32% have felt uncomfortable with AI-provided guidance, and 28% report engaging in harmful behaviors based on AI suggestions. The survey highlights a critical gap: while 61% use AI for mental health matters, only 38% trust AI platforms more than mental health professionals, suggesting users recognize AI's limitations while still relying on it as an accessible first line of support.
- Mental health is deteriorating worldwide, with 46% of survey respondents reporting they are struggling or languishing
- Users recognize both utility and risk: 55% are satisfied with AI advice, yet only 38% trust it more than professional mental health providers
Editorial Opinion
This survey captures a pivotal moment in mental health support where AI is becoming a primary resource not through clinical efficacy, but through accessibility and user desperation. While AI's 24/7 availability and zero cost make it invaluable for underserved populations, the data showing 28% of users experiencing harm highlights the urgency of developing safeguards, validation mechanisms, and clear limitations disclosure in mental health AI tools. As AI companies scale mental health features, they must balance accessibility with responsibility—not positioning AI as a replacement for professional care, but as a complement that recognizes its significant blind spots.
