A magic mushroom compound delivered in a converted community building has just shown real promise against depression that would not respond to anything else, NHS waiting times are edging down even as more people than ever seek help, and regulators are racing to catch up with the millions who have quietly started asking chatbots for therapy instead. Three different stories, moving at three different speeds, are shaping mental health care this year.

Mental health has never had a shortage of research interest, but it has often struggled to convert that interest into something patients can actually access. This year has produced a genuine clinical breakthrough for one of psychiatry’s hardest problems, alongside data showing a system handling record demand only in part, and a fast-moving public shift towards AI tools for emotional support that regulation has not yet caught up with. None of the three stories is finished. Together, they say a lot about where the field stands.

A Psychedelic Trial Built for the Real NHS

In August, researchers at King’s College London and South London and Maudsley NHS Foundation Trust published results from PsiDeR (Psilocybin in Depression Resistant to Standard Treatments), the UK’s first publicly funded, randomised controlled trial of psilocybin for depression delivered within an NHS setting. The results, published in Nature Medicine, come from 60 participants whose depression had not responded to at least two previous treatments. Half received a single 25mg dose of psilocybin, the active compound in magic mushrooms; half received a true placebo, with psychological support given to both groups throughout.

The difference between the two arms was substantial. By three weeks, 43% of the psilocybin group met the threshold for treatment response, rising to 50% by six weeks, compared with just 3% of the placebo group at both points. Remission rates followed a similar pattern: 40% of those given psilocybin no longer met clinical criteria for depression by three weeks, against 3% on placebo. Professor James Rucker, the trial’s lead investigator, said the study showed that “a randomised trial design of psilocybin compared to a true placebo was feasible in participants who had often failed many different types of treatment,” while cautioning that “individual responses were very variable and some of the difference seen between groups will be due to non-drug factors.” Most participants correctly guessed which group they were in, and the researchers note that expectation almost certainly accounts for some of the effect.

What sets PsiDeR apart from most psilocybin research to date is where it happened. Recruitment moved partway through the trial from a clinical research facility into a community mental health research building in a residential area, testing whether the treatment could be delivered safely outside a hospital. Catherine Bird, the trial’s senior clinical trials manager, said the move “showed us that psilocybin can be administered safely outside a hospital setting,” with no new safety concerns identified. Across both arms, 287 non-serious adverse events were recorded, the large majority mild, alongside four serious adverse events (three in the psilocybin group, one on placebo) that were not judged related to the drug. Professor Ben Carter, the trial’s lead clinical trial methodologist, called the results promising but stressed that “confirmatory trials are needed to demonstrate both the cost, and clinical effectiveness” before psilocybin could ever become available on the NHS.

The System Underneath Is Improving, but Only Some of It

Set against that research success is a more mixed picture of the service patients would actually rely on. The House of Commons Library’s most recent briefing, published in late July, shows that around 4.13 million people in England were in contact with NHS-funded secondary mental health services during 2024/25, just over 7% of the population and the highest proportion since records began in 2016/17. Waiting times have improved on paper: for the quarter ending May 2026, the median wait between referral and second contact was 37 days and the 90th percentile wait was 169 days, both down from 44 days and 221 days respectively a year earlier.

NHS Talking Therapies for Anxiety and Depression, the main route into treatment for common conditions, tells a similarly uneven story. In 2024/25, 1.81 million people were referred, 1.21 million entered treatment and 670,400 completed a course. The service beat its target for first treatment waiting times, with 91.6% of patients starting within six weeks against a 75% goal, but missed its target for second treatment, with only 82.7% starting within 18 weeks against a 95% goal. Just over half of patients who completed treatment moved into recovery, narrowly meeting the 50% target. Spending has grown considerably in real terms, up 31% since 2016/17 to a combined £20.6 billion planned for 2025/26, and integrated care boards are expected to meet the mental health investment standard requiring their mental health spending to rise at least as fast as their overall budgets. Children’s eating disorder services remain the weakest point in the system: only 73% of urgent referrals and 74% of routine referrals were seen within the NHS’s own target times in the most recent quarter, both well short of the 95% standard set a decade ago.

The Chatbot Problem Regulators Are Racing to Catch

The third story is the one moving fastest, and it is not coming from a lab or a hospital. A Mindometer survey by the British Association for Counselling and Psychotherapy (BACP) found that 64% of therapists reported a decline in public mental health over the past year, with 43% pointing to AI technologies as a contributing factor, and 28% saying they had seen clients receive unhelpful advice from tools such as ChatGPT. Among therapists working specifically with children and young people, 38% had observed a rise in children turning to AI chatbots for mental health guidance, and 19% reported cases where children received actively harmful advice. Separately, survey data cited by Mental Health UK suggests 37% of adults have used an AI chatbot for mental health or wellbeing support, and most of those used general-purpose tools rather than anything designed for the purpose.

Regulation is only now catching up. The MHRA is developing a new regulatory framework for AI in healthcare, informed by the National Commission into the Regulation of AI in Healthcare, with recommendations expected over the summer of 2026. In its response to the MHRA’s consultation, BACP argued that “AI is being introduced into healthcare at pace, but regulation has not yet caught up with the complexity of how these tools are being used,” and called for clearer definitions of what counts as AI in a healthcare context, along with shared accountability between developers, healthcare organisations and clinicians. Martin Bell, BACP’s Head of Policy and Public Affairs, put the appeal of these tools plainly: “It’s understandable that people are increasingly turning to AI for mental health advice because it’s available 24/7, feels non-judgemental, and offers a sense of privacy,” but added that “AI isn’t bound by ethical or confidentiality standards and, at the moment, lacks meaningful regulation and accountability.”

The government has moved on one narrow but symbolically significant part of the problem. In June, Prime Minister Keir Starmer and Technology Secretary Liz Kendall announced plans to restrict AI “romantic companion” chatbots, tools designed to simulate relationships or roleplay, to users aged 18 and over, alongside a wider ban on social media for under-16s. Legislation is expected before Parliament by the end of the year, with the first restrictions potentially in force by spring 2027. Critics have pointed out that the measure addresses only one narrow category of chatbot use and leaves the much larger pattern of children and adults turning to general-purpose AI for everyday emotional support largely untouched.

Three Speeds, One Field

What connects these three stories is a familiar tension: the science of treating mental illness is producing real, measurable progress, but the infrastructure meant to deliver that progress, and to manage the risks arriving alongside it, is not moving at the same pace. A rigorously tested psychedelic treatment is still years from any NHS patient being able to access it. A record number of people are reaching mental health services, but with some of those services, particularly for young people with eating disorders, still falling well short of the standards set for them a decade ago. And millions of people have already started using AI chatbots for emotional support that no regulator has yet properly defined, let alone overseen. Each strand is moving in the right direction. None of them is moving fast enough to say the job is done.

Sources include King’s College London, South London and Maudsley NHS Foundation Trust, Nature Medicine, the House of Commons Library, NHS England, the British Association for Counselling and Psychotherapy, Mental Health UK and the MHRA.