A nurse at Erlanger Baroness in Chattanooga admitted to stealing and abusing fentanyl for months, according to a nursing-board order reported by KFF/CBS News. The hospital uses Sentri7, an AI-powered drug diversion monitoring product from Wolters Kluwer, but the nursing-board order says Sentri7 failed to flag missing drugs and other "inconsistencies" that should have been detected. KFF/CBS reports that Erlanger declined to comment on Sentri7 or the diverted drugs, and a Wolters Kluwer spokesperson declined detailed answers while saying the company remains "confident in our software." The article notes that hospitals are not required to disclose use of this kind of software or to report malfunctions, and a Johns Hopkins neurologist and AI researcher told KFF/CBS that proprietary AI systems and limited transparency can allow errors to be buried rather than fixed.
What happened
KFF/CBS News reports that a nurse at Erlanger Baroness (also referred to as Erlanger Medical Center) admitted, in a nursing-board order, to pilfering and abusing fentanyl for months and was fired after failing a drug test. The nursing-board order states that Sentri7, an AI-powered drug diversion monitoring system from Wolters Kluwer, failed to flag missing drugs and other "inconsistencies" that the order says should have been flagged. Per KFF/CBS, Erlanger declined to comment on its use of Sentri7 or the diverted drugs. A Wolters Kluwer spokesperson declined to answer detailed questions but said the company remained "confident in our software."
Technical details
KFF/CBS reports that Sentri7 is marketed as an AI-based line of defense against drug diversion and that the Erlanger case provides a rare example where a nursing-board review found the system did not detect a prolonged pattern of theft. The article also reports that hospitals are not required to disclose their implementation of this software or to report malfunctions to regulators or the public.
Editorial analysis - technical context
Industry observers note that anomaly-detection systems for rare events routinely struggle when signal is sparse, labels are noisy, or contextual metadata are missing. Independent validation and transparent failure-mode reporting are recurring recommendations in the literature on clinical AI deployments.
Industry context
David Rastall, a Johns Hopkins Medicine neurologist and AI researcher, told KFF/CBS that because AI tools are often proprietary and hospital leaders may not fully understand their mechanics, a lack of transparency can permit errors to be repeated or buried rather than corrected.
What to watch
Observers will likely follow nursing-board proceedings, any vendor responses or post-incident audits, and whether regulators or hospital systems adopt disclosure or auditing practices for AI-based diversion monitoring.
Scoring Rationale #
The story highlights a practical safety failure of an AI tool in clinical operations, relevant to practitioners deploying anomaly detection in healthcare. It is notable but not paradigm-shifting, since it documents one reported failure rather than a systemic audit.
Practice with real Health & Insurance data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Health & Insurance problems