New warnings from The BMJ have intensified scrutiny of artificial intelligence in healthcare, with patient safety concerns now sitting alongside unresolved questions about liability, regulation and the pace of deployment across the NHS. In a recent BMJ news report, England’s patient safety commissioner said post-market surveillance of AI health tools must be strengthened to protect both clinicians and patients, while calling for clearer accountability when these systems cause harm. ([bmj.com](https://www.bmj.com/content/393/bmj-2026-100018?utm_source=openai))
Safety experts call for tighter oversight before wider rollout
The BMJ reported that AI is already being used in parts of the NHS for tasks including automated note taking, scan interpretation and triage, but that the regulatory framework has not yet caught up with the speed of adoption. Henrietta Hughes, who is deputy chair of the National Commission into the Regulation of AI in Healthcare, said surveillance after products reach the market needs to be “beefed up” so risks are identified earlier and more reliably. ([bmj.com](https://www.bmj.com/content/393/bmj-2026-100018?utm_source=openai))
The commission was established by the Medicines and Healthcare Products Regulatory Agency to help shape a new framework for AI medical devices, and its interim findings have already underlined the scale of the challenge. The BMJ’s reporting suggests the central concern is not whether AI will be used in clinical care, but whether the systems, oversight and responsibility structures are robust enough to keep pace with that reality. ([bmj.com](https://www.bmj.com/content/393/bmj-2026-100018?utm_source=openai))
Doctors face uncertainty over responsibility when tools fail
One of the clearest themes running through the BMJ coverage is the question of clinical liability. Hughes told The BMJ that clarity is needed on where responsibility sits if an AI tool harms a patient, a point that is becoming more urgent as the technology spreads into everyday hospital and primary care workflows. ([bmj.com](https://www.bmj.com/content/393/bmj-2026-100018?utm_source=openai))
Related BMJ reporting this week also highlighted broader professional unease about the direction of AI policy in the NHS, with doctors warning that the workforce plan is too reliant on the technology and should be paused. Taken together, the articles point to a growing divide between policy ambitions for efficiency and the practical demands of safe implementation. ([bmj.com](https://www.bmj.com/content/bmj/393/bmj-2026-240144.full.pdf?utm_source=openai))
Why the debate matters for the NHS
The issue matters because AI tools can influence diagnosis, documentation and patient flow at scale. If systems are introduced faster than governance can adapt, errors may be harder to detect and accountability harder to assign. The BMJ’s recent coverage suggests that many clinicians are not rejecting the technology outright, but are instead pressing for stronger safeguards before it becomes more deeply embedded in routine care. ([bmj.com](https://www.bmj.com/content/393/bmj-2026-100018?utm_source=openai))
For now, the message from the latest BMJ reporting is clear: AI may already be inside the NHS, but the rules governing its safe use are still being written. Until those rules are tightened, pressure is likely to grow on regulators, hospital leaders and policymakers to show exactly how patient safety will be protected in practice. ([bmj.com](https://www.bmj.com/content/393/bmj-2026-100018?utm_source=openai))
