In 2021, Johnson & Johnson’s Acclarent announced it had added artificial intelligence (AI) to its TruDi Navigation System, a medical device used to treat chronic sinusitis. This device, which had been on the market for three years, began receiving feedback from the FDA about malfunctions and injuries. After the AI integration, reports of issues jumped significantly, with at least 100 malfunctions and injuries noted by November 2025. At least ten people were reported injured, primarily due to incorrect information provided by the TruDi system during surgeries.
Some of the serious incidents included cerebrospinal fluid leaks, skull punctures, and strokes resulting from injuries to major arteries during procedures. While the FDA collects these reports, it does not determine the causes of such events, leaving the exact role of AI uncertain. Two stroke victims have filed lawsuits in Texas claiming that the AI in the TruDi system caused their injuries.
Johnson & Johnson referred further inquiries to Integra LifeSciences, which acquired Acclarent. Integra stated that the reports only showed that a TruDi system was used during surgeries where problems occurred and contended that there is no evidence linking the AI system to injuries.
As AI begins to change healthcare, many believe the technology will advance medical practices, but there have been concerns about safety. A review revealed that the FDA now authorizes over 1,357 AI-enhanced medical devices, which is double the amount authorized by the end of 2022. The TruDi system is not the only device questioned; the FDA has addressed reports about other AI devices, including heart monitors and ultrasound machines that failed to perform accurately.
Research from Johns Hopkins, Georgetown, and Yale found that 60 FDA-authorized medical devices using AI were involved in 182 recalls. Notably, 43% of these recalls occurred within a year of approval, which is notably higher than the typical recall rate for devices approved under similar FDA standards.
The rapid introduction of AI medical devices has challenged the FDA, as it works to manage the influx while dealing with losses in staffing. The U. S. Department of Health and Human Services is looking to increase its capacity to handle these approvals.
Additionally, generative AI chatbots are also emerging in the medical field, with many doctors using them to help with tasks like taking patient notes. However, patients’ use of AI for self-diagnosis could lead to complications.
AI gained significant attention following the launch of ChatGPT three years ago, which uses generative AI to create content. These AI tools have begun to be integrated into consumer healthcare applications. However, AI has been associated with medicine for decades, with its early usage in screening for cervical cancer starting in 1995. Current AI applications in medicine mainly utilize machine learning and deep learning to assist in diagnostic tasks and improve surgical procedures.
For example, Erin Ralph underwent a procedure using the TruDi system to treat her chronic sinusitis in June 2022. During the procedure, according to her lawsuit, the device misdirected the surgeon, resulting in injury to her carotid artery and causing a blood clot. After the surgery, Ralph suffered a stroke and required extensive recovery care, including a portion of her skull being removed to alleviate brain swelling. In an interview, she described ongoing challenges in her recovery.
Sinuplasty, a stroke, and a lawsuit
In May 2023, during a sinuplasty operation, patient Donna Fernihough suffered a stroke when her carotid artery was damaged, allegedly due to the TruDi system, which uses artificial intelligence. Fernihough filed a lawsuit against Acclarent and other manufacturers, claiming they knew the AI system could be inconsistent and unreliable. Acclarent denied these claims, stating it only distributed TruDi and did not design or manufacture it, while Integra LifeSciences, Acclarent’s owner, declared there was no evidence linking the AI to any injuries.
Dr. Dean, who has been consulting for Acclarent since 2014, received over $550,000 in fees, including at least $135,000 related to the TruDi system. His attorney stated that Dean could not comment due to patient privacy and ongoing litigation. After Integra acquired Acclarent, they indicated that Dean was no longer a consultant and that post-acquisition payments were for meals.
Reports from 2021 included a push by Acclarent’s then-president, Jeff Hopkins, to incorporate AI into TruDi for marketing purposes. The TruDi software is designed to assist in surgical planning and offer real-time feedback, but the lawsuit alleges that Acclarent rushed to market the technology without resolving safety issues, aiming for only 80% accuracy.
The FDA has noted that reports of device malfunctions are limited, often lacking details, and warns that these reports alone cannot assign blame. A review found at least 1,401 adverse event reports related to devices using AI from 2021 up to October 2025, with some documenting software issues. One notable report from June 2025 mentioned AI software, Sonio Detect, misidentifying fetal body parts during prenatal ultrasounds, but stated that no patient was harmed. Samsung Medison, the software’s owner, clarified that the report did not indicate any safety issues.
Additionally, there were 16 reports about AI-assisted heart monitors from Medtronic failing to detect abnormal heart rhythms, none of which reported injuries. Medtronic stated some incidents stemmed from user confusion. Their AI technology is designed to reduce false alerts but can misclassify actual heart events. Medtronic reviewed the cases and concluded only one event was missed, insisting that no harm resulted from these malfunctions, while also noting that some issues were associated with data display problems. The HHS did not address potential compliance matters related to these incidents.
FDA cutbacks under Trump
Five FDA scientists, both current and former, told Reuters that the agency is struggling to manage the increasing number of AI-powered medical devices. Approximately four years ago, the FDA added more scientists specializing in AI, particularly in the Division of Imaging, Diagnostics, and Software Reliability (DIDSR). This unit became essential for evaluating the safety of AI in medicine and grew to around 40 scientists by early last year. However, some senior regulators lack understanding of AI technology, leading to challenges in ensuring safety.
Recruiting skilled computer scientists for the government proved difficult, as many opted for higher-paying jobs in the private sector. These scientists worked to test AI device algorithms under different clinical situations and aimed to reduce instances of AI “hallucinations,” which are false outputs generated by the models.
In early last year, the Trump administration started cutting back on the AI team due to cost-cutting measures, resulting in around 15 scientists being laid off or leaving. The Digital Health Center of Excellence, which focused on AI policy, also lost about a third of its staff. Despite these reductions, an FDA spokesperson stated that rigorous standards are still applied to AI-related medical devices to ensure patient safety and promote public health.
Since the staffing cuts, some device reviewers have seen their workloads nearly double, increasing the likelihood of oversights. Unlike drugs, medical devices do not always require clinical trials before approval, and many AI-enabled devices can enter the market by referencing existing non-AI devices. Experts argue that the FDA’s traditional approach may not adequately ensure the safety of AI technologies, relying heavily on manufacturers to uphold quality standards.
With information from Reuters
