A nurse at Erlanger Baroness Hospital in Chattanooga, Tennessee, stole fentanyl over several months while artificial intelligence-powered drug diversion monitoring software failed to detect the theft, according to state nursing board records. The case exposes significant challenges in relying on AI systems to monitor controlled substance abuse in healthcare facilities.
What happened
In mid-2025, anesthesia staff at Erlanger Baroness noticed that nurse anesthetist John Stevenson appeared impaired, displaying slurred speech, lethargy, and difficulty staying awake while on duty in the surgery center. Following this, Stevenson failed a drug test and was terminated. He admitted to stealing and abusing leftover fentanyl from surgical procedures on a regularly increasing basis, sometimes daily, starting in March 2025. Tennessee’s Board of Nursing placed his license on probation and required drug counseling.
The hospital used Sentri7, an AI-powered medication-monitoring software by Wolters Kluwer designed to detect drug diversion faster than traditional methods. Despite this, Sentri7 failed to flag missing fentanyl doses and other irregularities linked to Stevenson’s actions. Hospital audits found multiple instances where the software did not raise alarms as expected during the months of diversion.
The Tennessee Department of Health disclosed the case in December 2025 through public disciplinary records. Erlanger and Wolters Kluwer declined detailed comment about the incident or technology performance.
Why it matters
Drug diversion—the theft or misuse of controlled substances in healthcare settings—is a widespread problem that endangers patient safety and public health. AI-based software like Sentri7 is increasingly deployed as a frontline defense to detect such theft through data analysis, but the Erlanger case raises concerns about their reliability and transparency. Failures in these systems can allow dangerous drugs like fentanyl, which is highly potent and linked to numerous overdoses, to be misappropriated for extended periods.
With hundreds of hospitals using similar AI monitoring tools and limited industry transparency, failures may be more common than publicly known. Experts warn that underreporting of AI malfunctions prevents improvements, and the complexity of operating rooms may inherently challenge AI tracking effectiveness. This case highlights the ongoing need for human oversight alongside technology in combating drug diversion.
Background
Drug diversion remains a persistent challenge in U.S. hospitals, with an estimated 15% of healthcare workers reported to divert medications at least once. Leftover drugs after surgery are common targets, and diversion can lead to medication shortages, contamination, and disease transmission. To combat this, hospitals have shifted from manual tracking to AI-augmented software systems such as Sentri7 and Bluesight’s ControlCheck, which monitor dispensing and waste documentation in real time.
A 2022 NIH-funded study found Sentri7 effective in detecting known diversion cases faster than human methods. However, real-world incidents like Erlanger’s reveal that technology alone may not fully prevent diversion, especially in complex environments like operating rooms where drug usage patterns differ. The combination of AI and vigilant human supervision is currently the most reliable approach to securing controlled substances in healthcare.
Sources
This article is based on reporting and publicly available information from the following source:
Read more Artificial Intelligence stories on Goka World News.