Between Innovation and Risk Aversion: A Contextual Perception of AI Sourcing Risks in Public Organisations
Year: 2026
Type: article
Abstract: Efforts to map and understand the risks associated with artificial intelligence (AI) are growing. While much of the existing discourse focuses on post-deployment risks, the system sourcing phase, where problems are defined and system requirements are specified, receives less attention. This paper explores AI sourcing risks, particularly those that arise early with cascading effects on system development. Drawing on interviews with 33 AI sourcing actors, we reveal procedural and conceptual risks associated with system sourcing and, in relation to these risks, conceptualise three interrelated risk constructs: pre-deployment operationalisation risks, risk-aware vendor risks, and adaptive future-proofing risks. The findings reveal how internal misalignments and vendor dynamics can compromise the success of AI projects. We propose a theory of contextual AI sourcing risk aversion to show the evaluative, context-dependent, and reflexive process in which actors construe and respond to risk under uncertainty, particularly in the context of AI sourcing in public organisations.
Source: AIS Electronic Library (AISeL)
Authors: Otunla, Zacchaeus, Mettler, Tobias
Institution: University of Lausanne
Language: English
Subfield: Management Information Systems
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7212342145
