The FDA has granted Breakthrough Device Designation to First Read, an AI tool built by Aidoc that reads chest X-rays and writes a full draft radiology report — not just a flag or a bounding box around a suspicious area, but actual report text describing findings, ready for a radiologist to review, edit and sign. The designation, announced June 25, 2026, is notable because it applies to a generative AI application in a part of medicine where regulators have historically been most comfortable with narrower, single-finding detection tools rather than open-ended text generation.
Breakthrough Device Designation doesn’t clear a product for market use — it’s a program that gives promising devices addressing serious conditions faster, more interactive engagement with FDA reviewers during development, aimed at shortening the path to an eventual clearance decision. Aidoc’s existing clinical AI platform, which flags abnormalities like pulmonary embolisms and intracranial hemorrhage for urgent radiologist review, is already deployed across nearly 2,000 hospitals and has analyzed more than 120 million patient cases, according to the company — the track record First Read is built on top of.
The Bottleneck First Read Is Meant to Attack
The tool exists because of a specific, well-documented problem: radiologist reporting turnaround times have been getting steadily worse. A study from the Neiman Health Policy Institute found that outpatient imaging interpretation turnaround times more than doubled between 2014 and 2023, with the steepest increases occurring in just the last two years of that window — a trend driven by imaging volumes rising faster than the radiologist workforce, particularly acute for routine studies like chest X-rays that make up a large share of daily reporting volume but individually carry lower reimbursement than more complex imaging.
First Read is designed to draft the routine, high-volume documentation work — describing what’s visible in a chest X-ray in structured report language — so that a radiologist’s time shifts toward reviewing, correcting and signing off, rather than composing reports from scratch for every study. Aidoc frames this the same way ambient AI scribes have been framed in outpatient medicine: not replacing clinical judgment, but taking over the transcription and drafting labor that surrounds it.
Why This Is a Bigger Regulatory Step Than It Looks
Reporting from STAT News in June 2026 noted that First Read was one of two generative AI radiology tools to receive FDA breakthrough designation nods around the same time — a signal that the agency is actively building a regulatory pathway for generative report-drafting tools specifically, rather than treating them as an edge case under existing detection-focused frameworks. That distinction matters: a model that only flags an abnormality has a narrower, more testable failure mode than one generating open-ended descriptive text, where an error could be a subtly wrong word choice buried in an otherwise plausible-sounding paragraph rather than an obvious miss.
The Case for Caution
Radiology safety researchers have raised a specific concern about generative report-drafting tools: automation bias, where a radiologist under time pressure may be more likely to accept an AI-drafted report with only a cursory review, especially if the draft reads fluently and confidently even when it contains a subtle error. Aidoc says First Read is built with safeguards specifically to reduce that risk, including maintaining mandatory clinician review and final sign-off on every report before it’s finalized, and the tool is explicitly positioned as draft-only — radiologists remain legally and clinically responsible for what the final report says. Whether those safeguards hold up against real-world time pressure, rather than in controlled validation studies, is a question the FDA’s ongoing review process is intended to probe further.
Following an Established Track
First Read builds directly on Aidoc’s existing FDA-cleared imaging triage tools, which the company argues gives it a validated technical foundation and clinical trust that a brand-new entrant wouldn’t have. That track record likely factored into the breakthrough designation, which tends to favor applicants who can point to substantial existing clinical deployment data over speculative new entrants.
What’s Next
Breakthrough designation typically shortens, but doesn’t guarantee, the path to eventual FDA marketing clearance — Aidoc will still need to complete the formal submission and clearance process, likely as a Class II device requiring clinician oversight, before First Read can be marketed for clinical use. If cleared, it would be among the most closely watched tests yet of how comfortable the FDA and hospital radiology departments are letting generative AI take the first pass at writing the words that go into a patient’s medical record.
