Abstract
This disclosure presents a Human-AI Trust Misalignment Detection System designed to identify and mitigate conditions in which human operators develop excessive or unwarranted reliance on AI generated recommendations within decision workflows. As artificial intelligence systems become increasingly embedded in operational, analytical, and supervisory environments, a critical socio technical vulnerability emerges when human oversight weakens due to overconfidence in automated outputs. Traditional AI safety efforts have largely focused on model robustness, accuracy, and infrastructure security; however, comparatively little attention has been given to monitoring the human behavioral response to AI recommendations in real time. The proposed framework introduces a behavioral analytics layer that continuously evaluates acceptance patterns, override frequency, response latency, and confidence divergence between human operators and AI outputs. By generating dynamic trust alignment scores and detecting early indicators of automation bias, the system enables organizations to intervene proactively before trust imbalance leads to operational, financial, safety, or security failures. The disclosed approach is broadly applicable across enterprise AI deployments, cybersecurity operations centers, financial decision systems, healthcare support platforms, and autonomous human-in-the-loop environments where maintaining calibrated human judgment remains critical.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Bhatnagar, Pranav Mr, "Human-AI Trust Misalignment Detection System: Behavioral Detection of Over-Reliance on Automated Recommendations", Technical Disclosure Commons, (February 23, 2026)
https://www.tdcommons.org/dpubs_series/9372