Overview
- The MHRA set out 44 formal recommendations from an independent commission that call for continuous post‑approval surveillance, powers to remove failing products, penalties for developers, trial 'L plate' supervision for new models, and patient rights to know when AI is used.
- The package is a set of policy proposals that will need government or parliamentary action and practical pilots before they become binding rules for developers and NHS services.
- Regulators say current device rules were built for static products like hip replacements and stethoscopes and cannot manage AI systems that learn, adapt and drift after authorization.
- AI tools are already in NHS practice, with reports that AI medical scribes are used by many GPs, and studies and clinicians warn that AI use can reduce patient disclosure and that doctors must still check and correct AI outputs.
- Supporters call AI a major clinical opportunity but caution about biased training data, incorrect or overconfident outputs, and the need for funding, technical capacity and clear legal responsibility before wider rollout.