Millions of Brits turning to AI chatbots for mental health support amid NHS challenges
More than 10 million Britons are now relying on artificial intelligence tools for mental health advice, as mounting pressure on NHS services pushes patients to seek support from chatbots like ChatGPT and Google Gemini, according to new research.
A study commissioned by cybersecurity firm NymVPN reveals that one in five UK adults (19%) are using AI chatbots to manage their mental health. Amid record demand and dwindling resources within the NHS, these tools are becoming a digital lifeline for those unable to access professional therapy.
In April alone, NHS England reported over 432,000 new referrals for mental health services, bringing the total number of patients receiving support to more than 2 million. With the Royal College of Psychiatrists forecasting a £300 million cut to mental health budgets over the next year, experts predict the trend toward AI-driven therapy will only accelerate.
But the shift raises serious questions about data privacy and digital safety.
A digital doctor in your pocket
Beyond mental health, nearly a third of the population (30%) are entering physical symptoms and medical histories into AI platforms to self-diagnose, blurring the line between convenience and clinical accuracy. The study also shows that millions are turning to AI for relationship advice, including help navigating break-ups and communication issues with partners.
“The likes of ChatGPT and Google Gemini are now so frequently used that they have ventured beyond being useful tools—they are being treated as therapists, doctors, and relationship coaches,” said Harry Halpin, CEO of NymVPN.
This comes as the NHS prepares to overhaul its 111 service with the integration of AI-powered triage, requiring people to adapt to a new digital-first healthcare landscape. However, the cybersecurity implications are giving some experts pause.
Caution amid convenience
NymVPN warns that while AI chatbots may offer easy access to support, they also pose a growing risk to user privacy. Nearly half (48%) of respondents described themselves as “cautious” about how much they reveal to these platforms, and 24% admitted they wouldn’t trust an AI-powered NHS service with their personal information, according to the research.
Halpin explains the dangers of sharing sensitive health information with AI platforms: “Health data is invaluable to cybercriminals. It can paint a detailed picture of someone’s vulnerabilities, making them a target for scams, blackmail, or malware attacks.”
He stressed that information shared with chatbots—such as names, detailed symptoms, or intimate feelings—could be stored, misused, or accessed by unauthorised third parties, particularly when privacy protections are weak or users share accounts.
Privacy best practices for AI users
NymVPN advises those turning to AI for health or relationship advice to take precautions:
1 - Avoid using real names or personally identifying details.
2 - Keep descriptions general rather than sharing specifics.
3 - Activate all available privacy settings.
4 - Consider using a VPN to protect location and device identity.
5 - Avoid account sharing, particularly on platforms that use conversation history to inform future responses.
“With NHS resources stretched to breaking point, the appeal of always-available AI tools is undeniable,” said Halpin. “But convenience should never come at the expense of safety. We need robust privacy protections in place.
As AI becomes increasingly embedded in daily life, experts are urging both users and policymakers to ensure that digital tools are regulated with the same rigour as any medical professional.