Community

NHS halts Foresight AI project amid patient data consent concerns

Sunday, 8 June 2025 1:08AM UTC

NHS England has paused its Foresight initiative, which used Meta’s Llama 2 AI to assess health risks from data on 57 million patients, after revelations about insufficient data permissions sparked concerns from medical bodies and highlighted public mistrust around AI in healthcare.

NHS England has announced a pause on a significant AI project aimed at assessing individual health risks due to emerging concerns regarding the use of patient data. The initiative, known as Foresight, was designed to utilise Meta's open-source AI model, Llama 2, in a bid to refine healthcare delivery by tailoring treatment plans based on comprehensive medical histories. Researchers from University College London and King's College London spearheaded the project as part of a national pilot exploring the potential of AI in personalising healthcare offerings.

However, the project faced scrutiny after it was revealed that data sourced from 57 million patients was being processed without appropriate permissions. Experts indicated that even anonymised data could potentially lead to the identification of individuals, a point underscored in reporting by The Observer. These concerns prompted the British Medical Association (BMA) and the Royal College of General Practitioners (RCGP) to call for a halt to the initiative. They raised alarms regarding data collected for COVID-19 research, emphasising that this had not been transparently conveyed to them before its utilisation in training the AI model.

Professor Kamila Hawthorne, chair of the RCGP, articulated that fostering patient trust is crucial in this context, asserting that medical data must not be used beyond the scope originally agreed upon by patients. She stated, “If we can’t foster this patient trust, then any advancements made in AI – which has potential to benefit patient care and alleviate GP workload – will be undermined.” The BMA echoed these sentiments through its England GP committee chair, Katie Bramall, who expressed surprise at the lack of awareness regarding the use of patient data for AI training.

The repeated emphasis on patient trust highlights a growing concern across the UK with regard to the use of AI in healthcare data analytics. A survey conducted in July 2023 revealed that 56% of UK citizens expressed a lack of trust in the NHS's handling of AI applications, stemming from security and privacy issues. Furthermore, 25% explicitly opposed the use of AI for processing their medical data, indicating a clear public apprehension about data privacy amid advancing technological integration.

While the Foresight initiative reflects broader trends towards harnessing AI in the NHS, it also underscores the delicate balance between innovation and the safeguarding of patient information. Similar initiatives have unfolded within the political sphere, as UK ministers have explored allowing private companies to profit from NHS data for AI advancements. This proposal faced immediate backlash over privacy concerns, highlighting the need for stringent controls and transparent governance in all AI-related endeavours in healthcare.

As a promising yet contentious field, the integration of AI in the NHS must navigate these waters carefully. Not only must it comply with ethical considerations, but it must also actively engage in transparent dialogue with patients to cultivate trust and maintain confidence in healthcare systems. The recent developments regarding the Foresight project serve as a pertinent reminder of these challenges, emphasizing that technological advancements should harmoniously coalesce with patient rights and expectations.

📌 Reference Map:

Paragraph 1 – ^[1], ^[4]
Paragraph 2 – ^[1], ^[2], ^[5]
Paragraph 3 – ^[3], ^[6]

Source: Noah Wire Services

More on this

https://www.dailymail.co.uk/news/article-14790993/NHS-biggest-AI-project-halted-57-million-permission.html?ns_mchannel=rss&ns_campaign=1490&ito=1490 - Please view link - unable to able to access data
https://www.theguardian.com/society/2025/jan/13/ministers-mull-allowing-private-firms-to-make-profit-from-nhs-data-in-ai-push - In January 2025, UK ministers considered allowing private companies to profit from NHS data to advance AI in healthcare. The plan aimed to use anonymised patient data for developing treatments and diagnostic tools. However, concerns arose regarding data privacy and public trust, as experts warned about potential misuse and the need for stringent controls. The initiative highlighted the delicate balance between fostering innovation and maintaining patient confidentiality in the NHS.
https://www.theguardian.com/politics/2025/jan/13/labour-ai-action-plan-nhs-patient-data-why-causing-concern - The UK's Labour Party's AI action plan, announced in January 2025, proposed leveraging NHS patient data to enhance healthcare through AI. While the plan aimed to improve services, it faced criticism over data privacy concerns. Experts highlighted the risks of re-identification from anonymised data and the importance of public trust. The initiative underscored the need for transparent data governance and ethical considerations in AI applications within the NHS.
https://www.ucl.ac.uk/news/2025/may/ai-model-trained-de-identified-data-57-million-people - In May 2025, researchers at University College London and King's College London initiated a pilot project to train an AI model named Foresight on de-identified NHS data from 57 million individuals. The model aimed to predict health outcomes and enable early interventions. Operating within the NHS England Secure Data Environment, the project sought to harness national-scale data to benefit diverse patient demographics, including minority groups and rare diseases.
https://www.infosecurity-magazine.com/news/uk-nhs-ai-privacy - A July 2023 survey revealed that 56% of UK citizens lacked trust in the NHS's use of AI for analysing patient data due to security and privacy concerns. Additionally, 25% of respondents opposed the NHS using AI to process their patient data. The findings highlighted the need for transparent communication and robust data protection measures to gain public confidence in AI applications within the NHS.
https://www.ft.com/content/9ec787a8-60d5-4899-8223-81335dfa919b - In January 2025, UK officials explored pricing structures for a proposed 'national health data service' as part of the government's 10-year NHS plan. This service aimed to centralise access to NHS patient data for companies and researchers. A government-backed review recommended creating transparent cost models to recover costs and value, rather than profiting from data. Concerns were raised about public mistrust if multinational companies profited from NHS data.
https://www.computing.co.uk/news/2464035/nhs-england-scraps-caredata-project - In 2014, NHS England discontinued the controversial care.data project following recommendations from the National Data Guardian for Health and Care, Fiona Caldicott, for better safeguards in sharing patient data. Despite the project's termination, the government and NHS England remained committed to data sharing to improve patient outcomes, indicating ongoing efforts to balance data utilisation with patient privacy.

Noah Fact Check Pro

The draft above was created using the information available at the time the story first emerged. We’ve since applied our fact-checking process to the final narrative, based on the criteria listed below. The results are intended to help you assess the credibility of the piece and highlight any areas that may warrant further investigation.

Freshness check

Score: 8

Notes: The narrative is recent, with the earliest known publication date being 7 May 2025. ([ucl.ac.uk](https://www.ucl.ac.uk/news/2025/may/ai-model-trained-de-identified-data-57-million-people?utm_source=openai)) However, similar concerns about AI models trained on NHS data have been reported since July 2023. ([infosecurity-magazine.com](https://www.infosecurity-magazine.com/news/uk-nhs-ai-privacy?utm_source=openai))

Quotes check

Score: 7

Notes: The quotes attributed to Professor Kamila Hawthorne and Katie Bramall are not found in the provided search results, suggesting they may be original or exclusive. However, without direct verification, this remains uncertain.

Source reliability

Score: 6

Notes: The narrative originates from the Daily Mail, a publication known for sensationalism and occasional inaccuracies. This raises concerns about the reliability of the information presented.

Plausability check

Score: 7

Notes: The claims about the Foresight AI project and the associated data privacy concerns are plausible and align with known issues regarding AI models trained on NHS data. ([ucl.ac.uk](https://www.ucl.ac.uk/news/2025/may/ai-model-trained-de-identified-data-57-million-people?utm_source=openai)) However, the lack of direct verification of quotes and the source's reliability affect the overall credibility.

Overall assessment

Verdict (FAIL, OPEN, PASS): OPEN

Confidence (LOW, MEDIUM, HIGH): MEDIUM

Summary: The narrative presents plausible claims about the Foresight AI project and associated data privacy concerns. However, the lack of direct verification of quotes and the source's reliability affect the overall credibility. Further verification from more reputable sources is recommended.

NHS
AI
Patient data
Privacy
Healthcare innovation