SHIFT Newsroom · UK
AI and mental health: what a UK survey can—and cannot—tell us
UK archive · Original report: 18 November 2025. Mental Health UK’s 2025 polling found both perceived benefits and reported harms. Self-reported experience is not proof of clinical effectiveness.
UK archive review · 13 September 2026. Original source: 18 November 2025. This article revisits a UK development; it is not a new announcement today.
A snapshot of reported use
Mental Health UK commissioned Censuswide to survey 2,000 UK adults aged 16 and over between 27 October and 3 November 2025. The charity reported that 37% had used an AI chatbot for mental health or wellbeing, combining occasional and regular use.
Respondents described accessibility and anonymity, alongside concerns about accuracy, privacy and harmful experiences. Reported use was higher among men than women. This does not establish why each person chose a chatbot or show that using one improved their health.
Benefit and safety need stronger tests
Feeling that a conversation helped is meaningful feedback, but it is different from a clinical outcome assessed in a controlled study. Likewise, a survey of reported harms cannot establish causation or compare the safety of individual products.
The charity called for evidence, transparency and human support. Those principles matter particularly when a tool sounds confident: fluency should not be mistaken for verification or clinical responsibility.
Questions worth asking about any digital support
Our suggested checks are who operates it, what it is intended to do, whether claims have been independently assessed and what happens to information you enter. Ask how to reach a person and what the service does when someone needs urgent help.
Do not use a chatbot response as the sole basis for changing treatment. For SHIFT, the editorial lesson is equally relevant: a polished article still needs traceable evidence. This survey is a reason to scrutinise digital support carefully, not an endorsement of a particular chatbot or a replacement for professional care.
Related UK reporting
Anxiety and NHS talking therapies: when to ask for support
Known facts
- Mental Health UK’s 2025 polling found both perceived benefits and reported harms. Self-reported experience is not proof of clinical effectiveness.