A study published in Radiology warns that AI-generated “deepfake” radiographs are realistic enough to mislead clinicians and even AI detection tools, reported Reuters (27 March 2026). Seventeen radiologists from 12 hospitals reviewed 264 X-rays, half of which were synthetic. Without prior warning, only 41% correctly identified fake images; accuracy rose to 75% after being told deepfakes were included. Researchers caution that convincing synthetic images could undermine trust in clinical evidence, enable fraud in litigation or insurance claims, and create cybersecurity risks if inserted into hospital systems. AI models also struggled, with detection accuracy ranging from 57% to 85%. The authors urge safeguards such as watermarking, secure image handling, and stronger audit trails.