Who is Danny / Shutterstock.com
Racial equity must be embedded across NHS artificial intelligence, the NHS Race and Health Observatory has said.
It makes the call in ‘Towards trustworthy, equitable AI in the NHS‘, a report published on 14 September 2026.
The report was produced following a roundtable held in July 2026 attended by organisations including the Department of Health and Social Care (DHSC), NHS England and the Patients Association.
In the report, the observatory says there is a “risk that AI can reproduce or amplify existing racial inequalities”.
“This can occur when systems are designed, implemented and governed without sufficient attention to ethnicity recording, fairness, inclusion and trust for diverse communities,” it adds.
It adds that “once trained, tested and deployed, AI systems can entrench, amplify and scale bias across every patient-facing decision they support”.
AI tools, it notes, can only perform according to the data on which they were trained and tested. The report says that to embed racial equity, tools must be “developed using diverse datasets that are representative of the populations a tool is intended to serve, alongside system-wide efforts to improve the routine collection and quality of ethnicity data within healthcare and research”.
Participants at the roundtable, the report adds, said AI developers and researchers should assess whether designs, models or datasets holds racist assumptions, or perpetuate race ideologies, rather than waiting to establish if bias emerges later.
It warns that the NHS has a “limited window” to embed racial equity across the lifecycle if AI, including in its design, procurement, implementation, monitoring and governance, “before inequitable practice becomes established and expensive to reverse”.
Habib Naqvi, chief executive of the NHS Race and Health Observatory, said: “AI has real potential to improve healthcare, but that potential will only be fully realised if these technologies are designed, regulated and deployed in ways that actively reduce inequalities rather than risk entrenching them.
“Our paper sets out why fairness, transparency, patient safety and public trust must be central to the development of AI, so that every community can benefit from innovation, and no one is left behind.”
Among its recommendations, the report says that anti-racism should be mandated in national requirements for AI, as procurement conditions rather than as guidance. It adds that published outcomes of NHS health inequalities should include the performance of AI-supported decisions around its performance by ethnic group.
Racially minoritised communities, it adds, should be involved in NHS AI at the procurement, evaluation and implementation stages.
Rachel Power, chief executive of the Patients Association, said: “Patients have told us that they want to be involved in decisions about how AI is used in their care and that involvement needs to happen from the start. This is particularly important for communities who may be more likely to experience bias, exclusion or barriers to accessing care.
“Patients should not be brought in once an AI system has already been designed. Their experiences and perspectives need to help shape how these systems are developed, tested, implemented and monitored.”
The NHS Race and Health Observatory report follows a government-commissioned report into the ‘Regulation of AI in healthcare‘. This said that “the future regulatory framework for software and AI-enabled medical devices needs to clearly address health equity as a critical aspect of safety and performance”.
“It is important these devices are developed, evaluated, deployed, and monitored with appropriate consideration of the diversity of the population, which may include characteristics such as age, gender, sex, race, ethnicity, and geographical location,” it adds.
The National Commission recommended that the Medicines and Healthcare products Regulatory Agency considers working with partners to provide guidance on health-equity expectations for manufacturers of AI-enabled medical devices, particularly around the intended population’s diversity and underserved population groups.
Responding to the NHS Race and Health Observatory’s report, Tase Oputu, president of the Royal College of Pharmacy, said: “AI could bring real benefits to healthcare but we need to make sure it doesn’t reinforce inequalities that already exist. Historic biases and gaps in health data can lead to poorer outcomes for racially minoritised communities if they are not actively addressed.
“To help ensure AI improves outcomes rather than widens inequalities, we need representative data, involvement from patients and pharmacy professionals in developing and testing AI, and checks that systems work fairly for different ethnic groups and communities.
“We also need clear accountability, ongoing monitoring and meaningful human oversight to make sure AI helps supports safe, equitable and person-centred care deliver safe and equitable care for everyone.”
