An NHS claim used to green-light a nationwide rollout of artificial intelligence has failed to meet official standards, the statistics watchdog has ruled.NHS England recently announced that an AI trial at a GP surgery had slashed phone queues by 29 per cent, a figure used to green-light a national AI rollout that will revolutionise the way millions of patients access NHS care.But the Office for Statistics Regulation (OSR) has now ruled that its use of the figure did not meet official standards because the data behind it was not published.The intervention is a major embarrassment for NHS England as it presses ahead with one of the biggest technological overhauls in the health service’s history.The AI triage tool is due to reach more than 200,000 patients within a year and eventually be available to every NHS App user by April 2028.It will question patients about their symptoms and help decide whether they should be directed towards a GP, pharmacy, A&E, community service or self-care.The rapid advance of AI into healthcare is already raising serious questions about patient safety and whether some of the most vulnerable could be left behind.Research shows that AI systems can produce inaccurate information, and researchers have warned of “automation bias”, where people become too reliant on a computer’s recommendations and fail to spot mistakes.Experts say an incorrect assessment could potentially mean symptoms are missed, a patient is directed to the wrong service or vital treatment is delayed.The NHS itself warns that organisations using some forms of AI technology may encounter inaccuracies that must be checked, while identifying privacy, information quality and automation bias among the risks.The rapid advance of AI into healthcare is already raising serious questions about patient safety and whether some of the most vulnerable could be left behind | PAThere are also fears that the drive towards an increasingly digital NHS could disadvantage elderly and vulnerable patients who struggle with smartphones, apps or technology.NHS England’s own patient safety framework says around 11 million people lack basic digital skills or do not use digital technology at all, and says they are more likely to be older and in poorer health.It specifically warns that increased use of digital healthcare risks widening inequalities for people including the over-75s if online platforms become the default way to access services.NHS England says patients will continue to be able to contact their GP through traditional routes alongside the new AI system.But the controversy over the evidence behind the rollout erupted after NHS England announced in July that the AI tool would be expanded following what it called a “successful trial”.It said: “An initial trial at a GP practice in Sussex resulted in a 29% reduction in the number of people queuing on the phone – helping to end the 8am rush while maintaining patient satisfaction levels.”Oxford academics Prof Carl Heneghan and Dr Tom Jefferson began investigating where the figure had come from, and published their findings on their Substack, Trust the Evidence.As previously revealed by GB News, they discovered that the scientific paper linked to NHS England’s announcement contained no analysis of phone queues, call waiting times or call volumes.Instead, the study compared the AI’s decisions with those made by GPs and examined patient satisfaction.Following questions from the academics and GB News, NHS England eventually clarified that the 29 per cent statistic came from a separate internal evaluation of telephone data.But the analysis had not been published, meaning outsiders could not examine the evidence behind a statistic being used to promote a national policy.Prof Heneghan and Dr Jefferson, from the University of Oxford’s Centre for Evidence-Based Medicine, complained to the statistics watchdog.Now the watchdog has found in their favour on the central transparency issue.Dr Tom Jefferson said: “The British public is expected to accept extraordinary claims from a press release from a public health authority for a programme which will be rolled out nationwide on the basis of unverifiable and inaccessible evidence. Why?”He added: “This is taxpayers’ money and this is people’s lives, and comes at a time when the public is losing faith in public health systems. Now the statistics watchdog agrees.”In an email seen by GB News, OSR head Ed Humpherson said: “We have found that NHS England’s statement about the GP practice trial did not meet our expectations” under official standards governing the public use of statistics and data.The regulator said public bodies should make data used in public statements available to everyone in a timely way, and that unpublished evidence should normally be released before or alongside the claim.The watchdog has raised the issue directly with NHS England | PAIt concluded: “The use of the 29 per cent figure in this NHS England statement does not align with these expectations.”The watchdog has raised the issue directly with NHS England.NHS England has now told the regulator it intends to publish the data behind the 29 per cent figure in the coming days, together with details of where it came from, how it was calculated and any important caveats.It is also taking steps to ensure “a similar issue does not happen again”, according to the OSR.