According to a ScienceDaily report published on March 26 2026, AI-generated deepfake X-rays have become realistic enough to fool both human doctors and existing AI detection systems. In tests, radiologists showed only limited success at identifying the fake images, with accuracy dropping further when they were not told they might be viewing fabricated scans. The ScienceDaily summary provides no details on study methodology, number of radiologists involved, sample size of real or fake images, or other experimental parameters. It is also unclear whether the underlying work is a peer-reviewed journal article or a preprint. These advances raise concerns about potential misuse including fraudulent medical claims and manipulated diagnoses. Experts cited in the report stress that stronger safeguards and improved detection tools are urgently needed as the technology improves. Source: https://www.sciencedaily.com/releases/2026/03/260326011452.htm. A key limitation is the absence of specific methodological information in the available release, preventing assessment of the tests' scale or rigor.