The medical and legal landscape in the United Kingdom saw warnings from the Medical Protection Society that doctors and NHS organisations could face negligence claims for errors made by artificial intelligence tools used in patient care. A report released earlier this week highlights a critical danger where clinicians risk becoming a “liability sink,” absorbing blame for technological failures that stem from the very infrastructure they rely on. This development transforms the conversation around IT maintenance from a technical necessity into a fundamental component of clinical risk management, particularly for organisations navigating the complex intersection of legacy systems and new AI applications.
The core of the issue lies in the interaction between outdated electronic patient-record systems and emerging AI diagnostics. The Medical Protection Society argues that unless the law is overhauled to protect medics, they will be held responsible for mistakes generated by software running on unstable or poorly maintained foundations. If an AI model provides a flawed diagnosis because it was fed stale or fragmented data from a neglected legacy database, the current legal interpretation suggests the clinician remains accountable. This reality was underscored by the Medicines and Healthcare products Regulatory Agency in its landmark public consultation report published on 11 June 2026, which stressed that “safe, operational AI must be reliably integrated into the NHS’s patchwork of legacy platforms to protect patients and avoid legal exposure.”
The tension was further inflamed when the British Medical Association branded a leaked government proposal to use AI as a primary substitute for staffing shortages as a “massive gamble.” This criticism reflects a deep-seated concern within the profession that deploying advanced tools without first securing the underlying digital estate invites disaster. The risk is not theoretical, it is a direct consequence of attempting to run high-speed algorithms on networks and databases that have been left to decay or operate without rigorous oversight. When an organisation fails to maintain the integrity of its older systems, it creates the precise conditions where AI errors become inevitable, yet the human operator bears the legal burden.
For healthcare providers, the solution is not to halt the adoption of AI but to recognise that the safety of their patients depends on the robustness of their entire technology stack, including the parts that are decades old. MicroSystem Support has long advocated for a hybrid approach where stable, legacy environments are maintained with the same rigour as modern infrastructure. By ensuring that older systems such as those running on DEC, OpenVMS, or Solaris architectures remain secure, patched and fully operational, organisations can prevent the data corruption that often triggers AI failures. This meticulous maintenance ensures that when a doctor reviews an AI suggestion, they are cross-referencing it against a source of truth that has been professionally safeguarded, thereby empowering them to exercise the human judgment required by the courts.
The path forward requires a change in how leaders view IT expenditure. Maintaining legacy systems is not just about keeping the lights on, it’s about creating a defensible position against future litigation. If a hospital can demonstrate that its foundational systems were managed to ISO9001 standards and protected under Cyber Essentials frameworks, it builds a case that the environment was safe for AI deployment. Neglecting these systems while chasing AI innovation leaves the organisation exposed to the exact scenario described by the Medical Protection Society, where the clinician is left as the sole target for systemic failure.
The recent news serves as a stark reminder that technology does not exist in a vacuum. The reliability of a new AI tool is inextricably linked to the health of the old system it connects to. As the MHRA notes, the goal is to ensure that new technologies deliver real-world benefit without compromising patient safety. Achieving this balance demands that we stop viewing legacy maintenance as a cost centre and start recognising it as the essential guardrail that makes AI adoption legally and clinically viable. In an era where the cost of error is measured in both lives and legal proceedings, the most prudent investment an organisation can make is in the expert stewardship of the data it has accumulated over decades.
References:
- Medical Protection Society (MPS): Report warning that doctors could face legal action over AI errors and become a “liability sink” (June 2026).
- Medicines and Healthcare products Regulatory Agency (MHRA): Public consultation report on safe AI integration in healthcare (11 June 2026).
- British Medical Association (BMA): Statement criticising the NHS AI staffing proposal as a “massive gamble” (June 2026).
- The Guardian: Coverage of the MPS report and the potential for negligence claims against doctors using AI tools.
