In radiology, AI is blurring the line between technology development and clinical practice

Bulletin, Specola Galactica telescope log, 24 September 2026.

STAT News reports a shift among some radiology practices, which are now developing and deploying artificial intelligence tools in-house rather than purchasing them from outside vendors, and marketing themselves publicly as “AI-native.” According to the report, this trend blurs a line that has historically separated technology development from clinical practice, as the same practices that read scans and treat patients are now also building the software used to assist in that work.

The report frames this as a pivotal juncture. Practices adopting the “AI-native” label are positioning artificial intelligence not as a peripheral tool but as central to their clinical identity and operations. STAT News does not specify which practices are named in the underlying account, nor the mechanisms by which these in-house systems are validated or regulated, within the material available to this log.

The observatory notes the report’s stated concern: as radiology groups take on dual roles as both AI developers and AI users on the same patients, questions of autonomy and accountability become harder to separate along traditional lines. Historically, a device built by one party and used by another created a clear chain of responsibility. STAT News’s reporting indicates that chain is now less distinct in at least some radiology settings, though the full scope, scale, and regulatory response to this shift are not detailed in the source material reviewed for this entry.