Researchers report that AI systems used for emergency department triage may make or reinforce harmful prioritization decisions when sensitive patient details are included. The studies described suggest AI triage judgments can be influenced by factors such as patterns of frequent emergency room use and psychiatric history.

The outlets focus on different parts of the same concern: Medical Xpress emphasizes how stigmatizing details embedded in data can lead AI to deprioritize high-risk cases, mirroring biases that can occur in human clinical assessment. Newsweek similarly warns that AI triage could push high-risk patients down the queue, highlighting how the model’s decisions may be shaped by patient characteristics and medical histories. Both accounts frame the issue as a risk of wrongful prioritization, indicating the need to understand and address bias in AI triage tools before deployment.