yougov.com via Reddit

YouGov poll: AI mental health fears climb to 43%

ai assistants ai ethics healthcare ai-mental-health polling consumer-trust

Key insights

  • 43% of Americans are very concerned AI will worsen mental health, up from 35% in June 2025, an 8-point rise in under a year.
  • Adults aged 45-64 saw the sharpest jump, moving from 37% to 50% concerned over the same period, crossing the majority threshold.
  • 37% of under-30s would be comfortable using an AI therapist, revealing a sharp generational divide in AI trust.

Why this matters

The 8-point jump in concern mirrors the pace of AI deployment in consumer mental health products, meaning platforms like Woebot, Wysa, and LLM-integrated therapy tools are entering a market where the majority of older demographic segments now view them with active suspicion. The under-30 cohort's openness to AI therapy creates clear product-market fit for B2C mental health AI, but regulatory bodies including state psychology boards and CMS are more likely to be shaped by the older, more skeptical majority. Founders building in this space face the specific challenge of designing for one generational cohort while navigating compliance and public opinion driven by another.

Summary

YouGov's survey of 2,180 US adults puts 43% now very concerned that AI will worsen mental health, up 8 points since June 2025. Adults aged 45-64 saw the steepest climb, jumping from 37% to 50% concerned over that same window. The generational inversion is the more striking finding. Despite rising systemic anxiety, 37% of under-30s say they'd be comfortable using an AI therapist, and a meaningful share believe they could form genuine emotional bonds with AI chatbots. Essentially: (YouGov) is mapping a fault line between fear of AI at the societal level and comfort with AI at the personal level. - Adults 45-64 crossed the majority-concern threshold for the first time, hitting 50%. - 37% of under-30s are open to AI therapy even as older cohorts grow more alarmed. - The gap between systemic fear and personal comfort is now the central tension in consumer AI adoption. This divergence will likely define the regulatory and product landscape for AI mental health tools over the next 12-18 months.

Potential risks and opportunities

Risks

  • AI-native mental health platforms (Woebot, Wysa, Character.ai) face accelerating backlash if high-profile adverse outcomes surface, with the 50%-concerned 45-64 cohort positioned to push state-level legislative action within 12-18 months.
  • Employers using AI mental health tools in EAP programs risk liability exposure as growing public concern translates into potential litigation or EEOC scrutiny over AI-mediated mental health support without adequate human oversight.
  • General-purpose LLM providers (OpenAI, Anthropic, Google) face reputational spillover if informal use of their products as mental health tools continues to grow without guardrails, directly amplifying the 43% concern figure in future polling cycles.

Opportunities

  • Hybrid AI-plus-human therapy platforms (Spring Health, Lyra Health) are positioned to capture users who want AI efficiency but are deterred by fully automated alternatives, as the 43% concern figure makes pure-AI therapy a harder sell to employers and insurers.
  • Compliance and safety tooling vendors focused on mental health AI can market directly to HR tech platforms and insurers now facing concrete liability questions about AI-mediated EAP tools, with the survey data as a ready-made business case.
  • Academic and nonprofit mental health organizations can establish AI safety benchmarks for mental health chatbots, gaining outsized policy influence as legislators in older-skewing electorates respond to the majority-concern numbers among 45-64-year-olds.

What we don't know yet

  • The survey does not specify which AI products respondents had in mind -- whether concern tracks dedicated therapy apps (Woebot, Replika) or general-purpose LLMs like ChatGPT used informally for mental health support.
  • No breakdown of under-30 respondents by prior AI chatbot use -- whether the 37% comfort rate is driven by actual users or hypothetical openness among non-users who haven't tried these tools.
  • Whether the June 2025 to May 2026 increase in concern correlates with specific adverse incidents (AI therapy app failures, chatbot-linked harm cases) or is purely driven by general AI exposure and media coverage.

Originally reported by yougov.com

Read the original article →

Original headline: YouGov: 43% of Americans Now Fear AI Will Worsen Mental Health — Up From 35% Last Year — While 37% of Young Adults Say They'd Comfort With AI Therapist