AI program overlooked evidence of harm in NHS maternity care

By the Editorial Team · Sunday, November 1, 2026

An AI program used to analyze over 10,500 submissions for a major maternity care review in England is being criticized for missing significant evidence of an NHS culture prioritizing vaginal births over patient safety. Critics argue this oversight, including concerns raised by a resigning obstetrician, has led to a report that underrepresents the impact of the "normal birth ideology," prompting a parliamentary inquiry into the AI's role in the review.

An artificial intelligence (AI) program used to analyze evidence submitted to a major review of maternity care in England has come under scrutiny for allegedly overlooking key concerns about the promotion of a “normal birth ideology” within the National Health Service (NHS). The review, led by Baroness Amos and published in June, examined more than 10,500 submissions from mothers and healthcare professionals, including detailed accounts of harm linked to NHS maternity practices.

Several mothers provided testimonies describing traumatic births in which their preferences for pain relief or caesarean sections were denied due to a perceived institutional emphasis on achieving vaginal deliveries without medical intervention. One focal point was Oxford University Hospitals Trust (OUHT), where 40 pages of evidence outlined cases in which women were encouraged to “trust their body” or pursue “normal birth at all costs.” These claims, however, did not figure prominently in the final report, which stated that there was no widespread evidence of such an ideology among NHS staff.

Baroness Amos later told the House of Commons health committee in July that the review team was unable to find evidence supporting the existence of a “normal birth at all costs” culture. Instead, the report identified failures in recognizing and escalating clinical problems and warned that individual women’s concerns were sometimes not fully acknowledged. This conclusion has been met with criticism from mothers and experts who argue that the report minimized the impact of normal birth ideology on patient safety.

Dr. Bill Kirkup, an obstetrician who resigned from the Amos inquiry, has raised concerns that the AI tool used to process the submissions failed to highlight evidence of normal birth ideology despite frequent mentions in interviews and panel discussions. Kirkup described the situation as a “conspiracy of silence” and has taken his concerns to the Commons health committee, which is now seeking clarification from Baroness Amos and Baroness Merron, the maternity minister, about how the AI was employed in the review.

Rebecca Matthews, a mother who experienced a traumatic birth at the John Radcliffe hospital in 2016, said the AI-assisted analysis contributed to ignoring crucial testimonies. Matthews, who suffered pre-eclampsia and was initially denied a caesarean despite clear medical risks, stated that in an NHS system prioritizing vaginal birth, safety concerns were sidelined. She received information through a freedom of information request that Microsoft’s Copilot AI software was used to analyze the submissions and produce thematic summaries for the review team. However, details of the prompts used with the AI remain undisclosed.

Historically, some NHS guidance encouraged vaginal deliveries without interventions such as pharmacological pain relief or forceps. The Royal College of Midwives (RCM) endorsed this approach until 2017, when it retracted the policy and issued an apology. NHS England also previously set targets limiting caesarean rates to around 20%, a measure discontinued in 2022 due to safety concerns.

Former Health Secretary Jeremy Hunt commented that if evidence regarding normal birth ideology was indeed excluded due to the AI analysis, it is deeply troubling and must be acknowledged and addressed to ensure improvements in maternity care. Baroness Amos did not respond to requests for comment on the allegations or the AI’s role in the review process.

This article was reported by synthesizing coverage from multiple independent news outlets to present a balanced, unbiased account. Read about our editorial standards. Spotted an error? Let us know.