AI earns clinical hours, raises new safety questions
Six stories: AI frees clinician hours, teens seek mental health from chatbots, and EU officials warn of AI-designed illegal drugs.
By Dr. Asher Knippel
Wednesday, 11 June: AI is saving clinicians more than 16 working days a year — but most say training lags behind
The eleventh annual Philips Future Health Index, released on 9 June and based on surveys of more than 2,000 healthcare professionals and 20,000 patients across ten countries, offers the most detailed picture yet of AI's day-to-day footprint in hospitals and clinics. According to the report, AI is now saving clinicians the equivalent of more than 16 working days per year. Nearly half (46 per cent) reported annual time savings of at least 132 hours, and 50 per cent said they can see an average of eight more patients each week as a result. Thirty-nine per cent said AI had identified or helped prevent a potential medical error at least three times in the preceding three months, while 60 per cent credited AI with improved precision in their work.
The picture is not entirely reassuring, however: 70 per cent of healthcare professionals described AI training as inadequate, inconsistent, or unavailable. The report cautions that health systems risk falling further behind unless they invest in structured, ongoing education alongside deployment. The survey covered Brazil, China, Germany, India, Japan, the Netherlands, Saudi Arabia, Singapore, the United Kingdom, and the United States.
Tuesday, 10 June: UK regulator opens world-first AI medicines safety 'sandbox'
The UK Medicines and Healthcare products Regulatory Agency (MHRA) announced on 9 June, during London Tech Week, the launch of a regulatory sandbox dedicated to testing artificial intelligence approaches to drug safety. The programme will provide a controlled environment to explore how AI can improve the accuracy of medicines assessments, better predict risks, and detect adverse effects that current pharmacovigilance methods miss.
Adverse drug reactions lead to approximately 250,000 hospital admissions in the UK each year, costing the National Health Service more than £2 billion. In the first phase, up to five AI-driven approaches will be tested starting summer 2026, including tools to predict how medicines behave across diverse patient groups — a response to the well-documented under-representation of women, older adults, and ethnic minorities in clinical trials. Although a UK programme, it may produce regulatory precedents relevant to the European Medicines Agency as both jurisdictions develop AI governance frameworks for post-market drug surveillance.
Tuesday, 10 June: 1 in 5 US adolescents turned to AI chatbots for mental health advice last year
A study published in JAMA Pediatrics (data collected November 2025, published June 2026) found that 19.2 per cent of US adolescents and young adults aged 12 to 21 — approximately 8.2 million people — used AI chatbots for mental health advice in 2025, up from 13.1 per cent in 2024. Among those users, 42.8 per cent did so at least monthly, and 91.7 per cent described the advice as somewhat or very helpful.
Most strikingly, 63.3 per cent of users had not disclosed their chatbot use to anyone — not a parent, therapist, or friend. The American Medical Association has called on Congress to establish guardrails, arguing that while AI chatbots may feel responsive and private, they carry no duty of care. The researchers note that access to AI mental health tools has grown fastest among teenagers in rural areas and low-income households where formal services are scarce — raising both access-equity and safety-equity concerns simultaneously. The study does not assess long-term outcomes. This dynamic resonates for Mediterranean readers: formal youth mental health services remain underfunded across Cyprus, Greece, and much of the EU periphery.
Tuesday, 10 June: UK invests £30 million to roll out AI cancer-diagnosis tools across the NHS
The UK government announced on 10 June a £30 million investment to accelerate AI-assisted cancer diagnosis across the National Health Service. The largest component — £20 million — will fund the roll-out of AI-powered chest X-ray analysis to every NHS trust in England by 2029; the technology has already helped more than four million patients receive faster lung cancer diagnoses or all-clears. A second tranche of £8.1 million will pilot six AI and digital technologies at 13 NHS sites, targeting faster pathways for heart failure, stroke, and lung cancer.
Data from 25 trusts already using AI-assisted X-ray review found that average reporting time for complex cases fell from eight days to four. The announcement forms part of the NHS's 10-Year Health Plan and is framed as a supply-side response to record diagnostic waiting lists. No randomised-controlled-trial data comparing AI-first versus standard pathways were cited in the government release, though lung-cancer AI tools have reported sensitivities exceeding 90 per cent in independent academic reviews.
Tuesday, 10 June: EU drugs agency warns AI is designing illegal psychoactive substances faster than bans can keep up
Dr Lorraine Nolan, Executive Director of the European Union Drugs Agency (EUDA), told the Financial Times on 9 June — ahead of the European Drug Report 2026 — that criminal networks are using the same generative AI tools employed in pharmaceutical research to engineer new psychoactive substances that fall outside existing banned-substance lists.
Approximately one new psychoactive substance was detected in Europe per week last year. AI accelerates this by enabling molecular-structure forecasting: a criminal actor can describe desired pharmacological effects and receive candidate molecules designed to mimic regulated substances while evading current legal definitions. "The same principles accelerating drug discovery in legitimate pharma are being applied on the illicit side," Dr Nolan said. The EUDA report is expected to recommend investment in AI-enabled early-threat detection for law enforcement. The problem is structurally difficult: enforcement operates on named-substance bans, while AI can generate an essentially unlimited sequence of structurally novel analogues.
Wednesday, 11 June: South Korea deploys AI companion dolls to counter elderly isolation
In South Korea — which has the world's fastest-ageing population — the government is supplementing formal elder care with AI-powered companion devices such as "Hyodol," reported by AFP on 11 June. These childlike interactive dolls speak, play music, remind users of medications, and flag signs of distress. Recipients include 78-year-old Bang Chun-ja, who lives alone and describes her Hyodol as her "only family member at home."
South Korea now has more people aged over 65 than under 15, and the ratio is projected to worsen sharply through the 2030s. The companion-doll programme is publicly funded in some municipalities, targeted at elderly people who live alone and score high on loneliness-screening tools. Early studies report reductions in loneliness and missed medications, but the evidence base is thin and long-term psychological outcomes have not been rigorously evaluated. The ethical question — whether AI companions supplement or substitute for human connection — remains unresolved and is increasingly urgent across the Mediterranean: Cyprus, Greece, and Italy face similar demographic pressures with fewer formal social-care options per elderly resident than northern European neighbours.
The content of this briefing is intended for general informational purposes and does not constitute medical advice. Readers should consult a qualified healthcare professional before making any change to their treatment, medication, or health regimen.