Health organisations raise concerns over AI-generated health information
A coalition of charities, including PIF along with Marie Curie, Cancer Research UK, Macmillan Cancer Support and the British Heart Foundation, has issued an open letter to the new Minister for Artificial Intelligence Kanishka Narayan.
115 health charities and membership organisations have signed the letter, which calls on the UK Government to lead the development of a UK-wide framework for safe AI-generated health information. The coalition seeks to ensure people are directed to trusted, evidence-based sources when seeking health advice online or using large language models (LLMs).
A previous survey of PIF members revealed that almost half of respondents had seen inaccuracies or misinformation about their topic area in Google’s AI overviews. The accuracy of AI-generated health information is increasingly important as fewer people then go on to access trusted online information and specialist support services.
Call to agree minimum safety standards
The coalition is calling on the UK Government to convene AI companies, the health system, regulators, patient organisations and the voluntary sector to agree minimum safety standards for AI-generated health information.
It wants a framework established to monitor and evaluate the impact of AI-generated health information on patient safety, public health outcomes and health inequalities.
In addition, the letter asks for:
- A recognised verification framework for trusted UK health information sources, building on initiatives such as the PIF TICK.
- AI systems operating in the UK to prioritise and reference verified NHS and UK-based accredited health information sources.
- Guidance on high-risk health topics where users should be directed to trusted information and support services.
- Mandatory transparency measures, including appropriate warnings where health information is AI-generated and may contain inaccuracies.
- Effective signposting to NHS services, healthcare professionals and specialist support organisations for information and support where appropriate.
Read the open letter in full here.
'Innovation should strengthen public access to trusted health information, not weaken it'
"Access to trusted, evidence-based health information is fundamental to helping people make informed decisions about their health and care. Artificial intelligence presents enormous opportunities, but it also introduces new risks when health information is inaccurate, lacks context or fails to direct people to trusted sources of support," says Sue Farrington, chair of trustees at PIF.
"As patient organisations, we are increasingly concerned that people may rely on AI-generated responses without understanding their limitations. We are calling on government, regulators, healthcare organisations and technology companies to work together to establish a framework that prioritises safety, transparency and accountability.
"Innovation should strengthen public access to trusted health information, not weaken it. Patients must remain at the centre of how these technologies are developed and deployed."
Dr Knut Schroeder, a practising GP based in Bristol, says: "When patients use AI for health advice, I frequently face the challenging task of debunking inaccurate information.
“Patients may delay seeking care for serious symptoms or arrive overly anxious because of an incorrect self-diagnosis. As a result, I spend valuable time untangling AI errors and rebuilding patient trust.”
'Information is often based on US sources and may be irrelevant in the UK'
Sophie Randall, director of PIF, adds: “Artificial intelligence can provide quick and accessible answers to people's health questions, particularly when they find it difficult to access the health service. But often the information provided is based on US sources and may be irrelevant in the UK.
“Ipsos data shows self reported public exposure to misinformation has increased fourfold since the introduction of AI. And 7 out of 10 of clinicians say they have had to correct AI-generated misinformation, placing further pressure on the NHS.
“1 in 5 of those who sought health advice from AI say the technology did not encourage them to seek medical support. It is in everyone’s interests to ensure the health advice published on AI tools is safe, accurate and includes onward signposting to further sources of support.”
Read the open letter, along with the supporting body of evidence, here. PIF members can also access a template letter to share with their supporters via the same link.