The Medicines and Healthcare products Regulatory Agency's (MHRA) National Commission into the Regulation of AI in Healthcare has published its summary of findings from its Call for Evidence. 761 individuals and organisations responded to include patients, clinicians, providers, industry, academics, and regulators.

The shared ambition was clear: people want AI to improve patient care, but they want confidence that these technologies are safe, effective, and properly governed.

The headline finding? Significant reform is needed but not a complete overhaul.

Ten key findings on regulatory gaps, liability, and AI oversight

The core criticism is that a regulatory framework designed for static medical devices is not well suited to iterative, adaptive AI systems. Ten findings stand out:

• Lifecycle-based, proportionate regulation: With stronger clinical evidence requirements and a risk-based approach that prioritises patient safety.
• Post-market surveillance must evolve: One-off approvals are insufficient; ongoing, real-world oversight is needed.
• Shared responsibility across the supply chain: Significant uncertainty persists as to where liability sits between providers, developers, and regulators.
• Human oversight must be retained: AI should augment, not replace, clinical decision-making.
• Transparency is non-negotiable: In procurement, regulatory documentation, and patient-facing communications.
• Data governance is unresolved: Patients want clarity on consent and commercial use; industry wants robust frameworks. Neither is satisfied.
• Incident reporting needs a rethink: Many clinicians are unaware the Yellow Card scheme already applies to AI-enabled medical devices. A centralised reporting system is widely supported.
• The liability framework is inadequate: Between 77% and 88% of respondents across all groups said the current legal framework is either insufficient or has significant gaps.

Why the liability point matters

As we have previously explored in AI in the healthcare sphere: Public and private considerations, the question of who bears responsibility when AI goes wrong in a clinical setting remains unresolved with key areas of uncertainty around the standard of care, informed consent, and policy silence on AI coverage. The MHRA Commission's findings confirm that this is not a niche concern: it is systemic.

Those issues are now being addressed at the highest levels and were considered by us in AI liability in healthcare: UK taskforce update. The UK Jurisdiction Taskforce, chaired by the Master of the Rolls, is shortly due to publish a Legal Statement on Liability for AI Harms. It has already confirmed that AI has no legal personality under the law of England and Wales. When read alongside the Commission's findings, the direction of travel is clear: reform is coming, and it will have real consequences for healthcare providers, insurers, and indemnifiers alike.

MHRA final recommendations expected summer 2026: What to prepare for

The MHRA Commission will publish its final recommendations during the summer of 2026, feeding into a new regulatory framework for AI in healthcare which is to be considered alongside the 10-Year Health Plan for England and the Life Sciences Sector Plan.

For those advising in this space, these findings are essential reading. The regulatory landscape is shifting, the legal questions are crystallising, and the time to prepare is now.

This article is part of a series on AI in healthcare and healthcare law by Naiomh O'Reilly, Associate at Browne Jacobson. See the previous editions of Insurance insights: Medical malpractice matters for more on how AI is shifting the legal sphere in medical malpractice.