The Medicines and Healthcare products Regulatory Agency (MHRA) has selected seven emerging artificial intelligence technologies for the second phase of its AI Airlock programme, which provides a controlled environment to test and refine AI medical tools before they reach patients.
The programme aims to help developers and regulators understand how AI technologies can be safely and effectively deployed in healthcare, identifying potential risks and pathways to regulatory approval. The selected tools include innovations that could reduce diagnostic times for bowel cancer, detect skin cancer and genetic eye diseases earlier, and improve hospital record-keeping and blood test interpretation.
The new cohort follows a successful pilot phase that involved companies such as Philips, Newton’s Tree, OncoFlow and Automedica. That initial phase helped identify regulatory improvements, including better methods for validating synthetic data, improving AI decision explainability, and addressing emerging issues such as AI “hallucinations”.
The MHRA describes the Airlock as a “regulatory sandbox” that allows manufacturers to test real-world AI healthcare applications in partnership with the regulator. The aim is to ensure that patient safety, data integrity and clinical reliability are maintained while supporting faster development and deployment of promising technologies.
Insights from the programme will inform the National Commission into the Regulation of AI in Healthcare, which brings together regulators, clinicians, patient representatives and technology companies to advise the MHRA on the future of AI oversight.
The seven manufacturers selected for phase two will test AI tools across a range of areas, from clinical note-taking to diagnostics and patient monitoring. Each project will undergo in-depth analysis to assess accuracy, bias, interpretability and regulatory compliance.
MHRA Chief Executive Lawrence Tallon said the UK was the first country to establish a dedicated regulatory sandbox for AI medical devices. “AI creates new opportunities and new complexities,” he said. “By working directly with innovators we can anticipate regulatory challenges, support responsible innovation and protect patients.”
The MHRA has also published four reports from the first phase of the Airlock, outlining lessons learned and recommendations for improving regulation of AI medical devices. The documents provide a framework for how testing, data validation and ongoing monitoring can be integrated into future regulatory processes.
Phase two of the Airlock will run through 2026, with findings used to shape future guidance for AI technologies seeking approval under UK medical device regulations.
The MHRA said it would continue working with industry and clinical partners to ensure that safe, evidence-based AI systems can reach the NHS and support the government’s goal of creating the world’s most AI-enabled healthcare system.