AI health tools being introduced without liability concerns addressed, report warns

The rapid rise of artificial intelligence (AI) in healthcare is happening without the basic legal safety nets needed to protect patients and health workers, the World Health Organisation (WHO) has warned.

A detailed report looking at the use of AI across European countries, including the UK, found widespread use in helping doctors spot diseases, reduce administrative tasks and communicate with patients.

But questions need to be answered around who is responsible when an AI system makes a mistake or causes harm, the report said.

While nearly all countries in the European Region have recognised the potential of AI to transform healthcare, ‘readiness’ remains uneven and fragmented, the report found.

Only four countries of 50 European countries looked at have a dedicated national AI strategy for health, with a further seven developing one, it found.

Across Europe, certain areas of data governance are still lagging, including guidance on the secondary use of health data for public-interest research, rules for cross-border data sharing and frameworks for collaboration with private companies on public-interest health research, WHO said.

Without addressing these gaps, AI initiatives risk producing technically advanced tools that do not fully meet clinical or public health needs, the report concluded.

Dr Hans Henri P Kluge, WHO regional director for Europe, said: ‘AI is already a reality for millions of health workers and patients across the European Region.’

‘But without clear strategies, data privacy, legal guardrails and investment in AI literacy, we risk deepening inequities rather than reducing them.’

The report welcomed some proactive steps taken by some countries giving the example of Estonia which links electronic health records, insurance data and population databases into a unified platform that now supports AI tools, and investment in Finland into AI training for health workers.

It also noted that Spain was piloting AI for early disease detection in primary health care.

The UK Government is keen to embrace the use of AI in healthcare to drive efficiencies and has announced the roll-out of tools across GP practices.

But across Europe, regulation is struggling to keep pace with technology, it found.

Nearly 9 out of 10 countries said legal uncertainty was the primary barrier to AI adoption and 78% cited financial constraints as a major obstacle.

The UK response placed infrastructure at the top of the list of barriers, followed by strategy, data quality and standards, financial constraints, capacity, trust and environmental impact.

Half of countries had chatbots in place for patient assistance, which the UK cited as an already ‘established’ use of AI.

Meanwhile, less than 1 in 10 countries have liability standards for AI in health, which determine who is responsible if an AI system makes an error or causes harm.

Dr Natasha Azzopardi-Muscat, director of health systems for WHO/Europe, said: ‘We stand at a fork in the road. Either AI will be used to improve people’s health and wellbeing, reduce the burden on our exhausted health workers and bring down healthcare costs, or it could undermine patient safety, compromise privacy and entrench inequalities in care. The choice is ours.’

WHO called on countries to clarify accountability, establish redress mechanisms for harm, and ensure that AI systems ‘are tested for safety, fairness and real-world effectiveness before they reach patients’.

The report comes as the British Association for Counselling and Psychotherapy warned of the growing risks of children using AI tools such as ChatGPT for mental health advice. 

A BACP survey found more than a third (38%) of therapists working with under 18s have clients seeking mental health guidance from AI platforms and one in five reported children receiving harmful mental health advice. 

Therapists said some AI tools are providing potentially harmful and misleading information – including encouraging children to self-diagnose conditions such as ADHD and OCD, reinforcing avoidance behaviours and automatically validating their feelings regardless of what they express. 

BACP said it was particularly concerned about AI’s inability to offer real-time support or intervene in crisis situations.