The NHS has suspended its Foresight AI programme after concerns emerged over the use of anonymised health records from 57 million patients without adequate medical consultation, sparking debate on data privacy and trust in AI-driven healthcare.
One of the most ambitious artificial intelligence projects within the NHS has come to a halt, following significant concerns that it may have improperly utilised the health records of approximately 57 million individuals. The programme, known as Foresight, employed Meta’s open-source AI model, Llama 2, and aimed to predict future medical events by analysing past patient data. While the data used was reportedly stripped of personal identifiers and addresses to maintain confidentiality, experts have cautioned that even anonymised health records can potentially be reconstructed to identify individuals. Importantly, Meta, the creator of the AI model, did not have access to any patient data.
The Foresight project was approved through a fast-tracked process implemented during the COVID-19 pandemic, aimed primarily at facilitating urgent research related to the virus. However, some researchers have raised concerns regarding the appropriateness of this approach, questioning whether the project’s goals genuinely aligned with pandemic-related research. One participant noted the lack of clarity regarding how the AI programme contributes to our understanding of COVID-19, suggesting that the approval may have been inappropriately justified.
Both the Royal College of General Practitioners (RCGP) and the British Medical Association (BMA) have expressed alarm over the apparent lack of consultation with medical professionals before the data was shared with Foresight. They argue that such actions could undermine public trust in the NHS and erode confidence in the deployment of AI technologies for healthcare. In a statement, Professor Kamila Hawthorne, chair of the RCGP council, emphasised the need for patient trust, saying, “Patients need to be able to trust their personal medical data is not being used beyond what they’ve given permission for.” This sentiment underscores the ethical imperative for respecting patient data privacy, especially in light of the potential benefits of AI in alleviating pressures on healthcare systems.
An NHS spokesperson confirmed the suspension of the Foresight research, acknowledging that while doctors may review the general data-sharing agreement tied to the model, they were not specifically consulted concerning this project. This situation has rekindled wider debates about the management of patient data, especially in the context of the UK government’s ongoing AI initiatives which aim to use NHS data to encourage healthcare innovation.
Labour’s proposed AI action plan includes the establishment of a National Data Library designed to facilitate tech startups and researchers in training new models using NHS data. While the overarching goal is to improve healthcare, substantial apprehensions regarding patient privacy and data security persist. Experts warn that anonmymised datasets are not infallible, and any compromise in data security could lead to detrimental breaches of patient confidentiality. Additionally, this plan contemplates the prospect of private companies profiting from NHS data, further complicating the ethical landscape surrounding data usage and public trust.
The government’s consideration of allowing private firms access to NHS patient data is framed as a strategy to propel AI-driven healthcare advancements. However, this prospect raises fundamental questions about the balance between innovation and privacy. Experts advocate for stringent oversight to ensure that patient data remains protected from misuse, highlighting the necessity for transparent policies in any engagement with AI applications in health settings.
This evolving scenario encapsulates the intersection of healthcare, technology, and ethics—underlining the critical need for maintaining public trust in the NHS, particularly as artificial intelligence increasingly shapes the landscape of patient care. As discussions progress, the importance of patient consent and the integrity of personal medical data must remain at the forefront of healthcare innovation initiatives.
Reference Map:
- Paragraph 1 – [1], [4]
- Paragraph 2 – [1], [2], [5]
- Paragraph 3 – [3], [6]
- Paragraph 4 – [5], [7]
- Paragraph 5 – [2], [6]
Source: Noah Wire Services
- https://observer.co.uk/news/national/article/nhs-england-pauses-ai-project-fed-health-data-of-57m-patients – Please view link – unable to able to access data
- https://www.theguardian.com/politics/2025/jan/13/labour-ai-action-plan-nhs-patient-data-why-causing-concern – Labour’s AI action plan proposes creating a National Data Library to enable tech startups and researchers to train new models using NHS data. While aiming to improve healthcare, concerns arise over patient privacy and data security. Experts warn that even anonymised health data can be re-identified, potentially compromising individual privacy. The plan also considers allowing private companies to profit from NHS data, raising ethical questions about data usage and public trust. The Royal College of GPs and the British Medical Association have expressed apprehension over the plan’s implications for patient confidentiality and consent.
- https://www.theguardian.com/society/2025/jan/13/ministers-mull-allowing-private-firms-to-make-profit-from-nhs-data-in-ai-push – UK ministers are contemplating permitting private companies to profit from NHS data to advance artificial intelligence in healthcare. The proposal includes using anonymised patient data to develop new treatments and diagnostic tools. While aiming to boost economic growth and healthcare innovation, this initiative raises concerns about patient privacy, data security, and the potential for misuse. Experts highlight the need for stringent controls to protect patient data and maintain public trust. The plan has sparked debate over the balance between technological advancement and ethical considerations in data usage.
- https://www.completeaitraining.com/news/ai-model-trained-on-57-million-nhs-records-sparks-privacy/ – The Foresight AI model, trained on de-identified data from 57 million NHS patients, aims to predict health outcomes and assist in early interventions. However, experts express concerns about the potential for re-identification of individuals, even with anonymised data. The project has sparked debates over data privacy, consent, and the ethical use of health information. While the model could revolutionise patient care, it underscores the need for robust data protection measures and transparent policies to maintain public trust in the use of health data for AI research.
- https://www.the-independent.com/news/uk/politics/artificial-intelligence-nhs-patients-medical-records-data-sharing-google-deepmind-a8000856.html – The UK government’s AI action plan proposes sharing anonymised NHS patient data to foster innovation in healthcare. While aiming to enhance medical research and treatment development, this approach raises concerns about data privacy and patient consent. Previous initiatives, such as the care.data programme, faced public backlash over data-sharing practices. Experts emphasise the need for transparent communication and robust safeguards to protect patient information and maintain public trust in the NHS’s data-sharing practices for AI research.
- https://www.techxplore.com/news/2025-02-uk-ai-health.html – The UK government’s plan to utilise public data, including NHS health records, for artificial intelligence research has sparked concerns over patient privacy and data security. While aiming to drive innovation, experts warn about the risks of re-identification and potential misuse of health data. The initiative highlights the need for stringent controls and transparent policies to protect patient information and maintain public trust in the use of health data for AI research.
- https://www.digitalhealthnews.com/uk-govt-eyes-nhs-patient-data-for-ai-research-amid-privacy-concerns – The UK government is considering allowing private companies to access NHS patient data for AI research, aiming to drive innovation in healthcare. However, this proposal has raised privacy concerns, with experts warning about the potential for re-identification of individuals even with anonymised data. The initiative underscores the need for robust data protection measures and transparent policies to maintain public trust in the use of health data for AI research.
