A broad review of artificial intelligence-enabled medical devices cleared by the FDA found that most were not tested to determine whether they improve patient outcomes. Researchers analyzed 1,357 AI-enabled devices authorized by the agency through December 5, 2025, and reported that only a minority had evidence tying their performance to health improvements in real-world settings. The review underscores a recurring issue for the AI medical device sector: regulatory clearance can be based on performance characteristics without a structured outcomes study. For sponsors and developers, the results spotlight the need for more outcome-driven evidence generation, including prospective evaluation designs. The finding is likely to intensify expectations from payers, health systems, and regulators for post-clearance evidence on clinical utility.