A sweeping review of AI-enabled medical devices authorized by the FDA found that most were cleared without evidence tied to improved patient outcomes. Using a dataset of 1,357 AI-enabled devices cleared through Dec. 5, 2025, researchers reported that only a small subset had been tested to show clinical benefit. The review highlights a mismatch between device authorization and outcome demonstration, emphasizing that many clearances rely on performance metrics rather than endpoints linked to health improvements. The report frames this as an issue for patient safety, comparative effectiveness, and the credibility of AI claims when translational evidence is thin. The findings arrive as companies pursue expanding indications and as regulators consider frameworks for evaluating generative AI-enabled medical devices, increasing pressure for clearer evidentiary standards across the AI device lifecycle.