# UK Needs New Laws for AI in Healthcare, Says Watchdog
The UK's medicines regulator is calling for urgent legislative changes to govern artificial intelligence in the NHS. Lawrence Tallon, chief executive of the Medicines and Healthcare Products Regulatory Authority (MHRA), told the BBC that AI deployment in the health service is accelerating faster than existing laws can accommodate.
Tallon's warning comes as NHS trusts and hospitals begin integrating AI tools into clinical workflows. The technology is already being piloted for diagnostic imaging analysis, administrative tasks, and clinical decision support. Without new regulation, Tallon argues, there is insufficient oversight of how these systems perform, who bears responsibility when they fail, and how patient safety gets protected.
The current regulatory framework treats AI as a software tool or medical device, applying existing rules designed for traditional technologies. This approach creates blind spots. AI systems learn and evolve over time, making their behavior unpredictable in ways that static devices never are. They can develop biases that disproportionately harm certain patient populations. They can fail silently, providing incorrect recommendations while appearing confident. Traditional regulatory pathways don't adequately address these risks.
Healthcare AI systems already in use include algorithms that flag patients at high risk for sepsis, tools that help radiologists detect cancers from imaging scans, and systems that optimize hospital bed allocation. These applications offer genuine benefits. Earlier detection saves lives. Better resource allocation reduces wait times. But each deployment carries risks that current rules don't fully capture.
The MHRA has begun working with other regulators and NHS leadership to identify gaps. Tallon emphasized that new legislation should clarify accountability. When an AI system makes an error, clinicians need to know whether they should trust the recommendation, how to verify it, and who is liable if harm occurs. Healthcare workers also need training on AI limitations and proper use. Hospitals need audit trails showing which systems influenced which decisions.
The regulator is also concerned about the pace of change outstripping scrutiny. New AI models are released constantly. The NHS lacks systematic processes for evaluating whether new tools meet safety and effectiveness standards before deployment. Some trusts have adopted commercial AI products with limited evidence of benefit in real-world NHS settings.
Tallon's position reflects growing concern among medical regulators globally. The US Food and Drug Administration has already issued guidance for AI in healthcare. The European Union is developing AI Act provisions specifically for medical applications. The UK risks falling behind without comparable frameworks.
The call for new laws faces practical challenges. Legislation takes time to draft, debate, and implement. The NHS moves quickly when it identifies promising technologies. Regulators must balance moving fast enough to prevent harmful deployments while not moving so slowly that they block beneficial innovations.
Healthcare professionals themselves remain divided. Some see AI as essential for addressing NHS staffing shortages and diagnostic backlogs. Others worry about losing clinical judgment and accountability. Robust regulation could bridge this gap by ensuring AI augments rather than replaces human expertise and by creating clear lines of responsibility.
The MHRA's position signals that voluntary adoption of AI without legal guardrails will not continue indefinitely. New rules appear inevitable. The question now centers on what those rules should require to protect patients while allowing the technology to deliver on its genuine promise.