Tag
The article proposes a framework for quantifying system-level harms from AI adoption in complex sociotechnical systems, using a case study on adversarial manipulation in financial infrastructure to demonstrate increased risks.
This paper presents a qualitative case study of a large IT services company's 2025 development and rollout of an agentic AI system, distilling seven lessons for embedding governance into system architecture and operations to balance autonomy with accountability.