AI Governance Lacks Accountability in Automated Decisions

Sopra Steria Next, in its latest CIO Compass, highlights a crucial gap in AI governance. Despite years focusing on data, large organizations have not prioritized governing AI decision-making. Only fewer than 20% of AI-influenced decisions have a designated business owner, explicit rules, and accountability.

Furthermore, less than a quarter of AI-driven decisions in production can be traced end-to-end. This lack of oversight poses risks, as automated decision failures can impact significantly more operations and revenue compared to traditional IT incidents.

The report calls for a broadened understanding of AI sovereignty. It stresses the need for a 'sovereignty stack' covering data, models, platforms, APIs, and decisions. This is necessary to determine who operates and is accountable for these layers.