Healthcare AI is moving faster than the evidence to trust it.We track the safety gap.
npj Digital Medicine: in a multicentre study across Chinese hospitals, junior clinicians reviewing GPT-4o's output in simulated clinical scenarios identified only 15.8% of its hallucinations.
Nature Medicine: an explainable AI raised physicians' accuracy at predicting immunotherapy response in lung cancer from 57% to 65%, but lung-cancer experts followed its incorrect suggestions 72.2% of the time.
The MHRA's independent National Commission, after a year and more than 12,000 contributors, recommends staged 'L-plate' authorisation for new healthcare AI, continuous real-world monitoring for a device's whole working life, and a public, searchable record of AI safety incidents.
A living record of the research, reports and news shaping AI safety in healthcare — so you can adopt with your eyes open, while the frameworks catch up.
Free to join. Unlock the complete, weekly-updated archive of 51 sources — sortable, with the monthly briefing.
A synthesis of public, peer-reviewed and official sources, maintained by Coherent Healthcare. Every claim is cited; verify at source before external use. Commercial surveys and unproven litigation are flagged as such; "risk" here means published concern, not proof any product is unsafe. Not medical, legal or procurement advice.