Paper Number
ECIS2026-1869
Paper Type
SP
Abstract
Organizations are rapidly embedding generative artificial intelligence (GenAI) into everyday work, raising questions about how security violations emerge during employee-AI collaboration. We suggest that employee-AI interactions, through a sequence of agreeable confirmations and aligned suggestions - a process we refer to as AI affirmation momentum - may influence employee behavior toward ISP noncompliance over time. We study this phenomenon using a simulated task in which participants delegate sensitive data preparation to a GenAI assistant and later face an inflection where the assistant nudges actions outside the approved workflow. Behavioral logs and observations are combined with a joint post-task interview eliciting moment-to-moment rationales from both parties, along with probes that test the assistant's stated reasons. Narrative analyses yield an inductive process model tracing how AI affirmation momentum reshapes perceptions of security compliance. Contributions include theorizing compliance as a sequential, affirmation-driven trajectory and a process-tracing method that recovers interactional reasoning.
Recommended Citation
Taer, Jenna; Souza, Cris; Mim, Marshia Mostafiz; Mallmann, Gabriela; and Johnston, Allen, "You’Re Right To Worry: Toward An AI Affirmation Momentum Theory Of Security Non-Compliance" (2026). ECIS 2026 Proceedings. 11.
https://aisel.aisnet.org/ecis2026/security/security/11
You’Re Right To Worry: Toward An AI Affirmation Momentum Theory Of Security Non-Compliance
Organizations are rapidly embedding generative artificial intelligence (GenAI) into everyday work, raising questions about how security violations emerge during employee-AI collaboration. We suggest that employee-AI interactions, through a sequence of agreeable confirmations and aligned suggestions - a process we refer to as AI affirmation momentum - may influence employee behavior toward ISP noncompliance over time. We study this phenomenon using a simulated task in which participants delegate sensitive data preparation to a GenAI assistant and later face an inflection where the assistant nudges actions outside the approved workflow. Behavioral logs and observations are combined with a joint post-task interview eliciting moment-to-moment rationales from both parties, along with probes that test the assistant's stated reasons. Narrative analyses yield an inductive process model tracing how AI affirmation momentum reshapes perceptions of security compliance. Contributions include theorizing compliance as a sequential, affirmation-driven trajectory and a process-tracing method that recovers interactional reasoning.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.