Sopra Steria Next: Less than 25% of AI Decisions End-to-End Traceable
In the latest installment of its CIO Compass series, published on October 20, 2026 under the title "Without Control or Purpose, AI is Worthless", Sopra Steria's consulting firm argued that AI sovereignty is not limited to data and model localization. According to the document, it also encompasses control of automated decisions and their consequences.
Less than a Quarter of AI-Driven Decisions End-to-End Traceable
The report, presented as an analytical framework for Chief Information Officers, relies on several quantified findings. Less than 20% of decisions influenced or executed by AI have an identified business owner, explicit rules, and formalized accountability. Less than 25% of AI-driven decisions in production can be traced and replayed end-to-end, and a failure in automated decision-making can affect 10 to 100 times more operations, customers, or revenue than a standard IT incident. Sopra Steria Next notes that large organizations have spent ten years governing their data while giving little or no thought to decision-making governance, whereas generative and agentic AI "prioritizes, triggers, and executes".
The firm believes that vendor dependency "has shifted from infrastructure to cognition". It urges Chief Information Officers to make explicit a five-level "sovereignty stack" (data, models, platforms, APIs, and decisions), accompanied by three questions: who operates, who can replace, who is responsible. "A system you cannot stop in real time is not governed: it is observed", stated Valérie Thiblet, Partner Data & AI at Sopra Steria Next.
Nine Dimensions to Evaluate Data & AI Strategy and Creation of a New Function
The document proposes an analytical framework across nine dimensions to assess the current state and build a Data & AI strategy, from pilot to deployment. It identifies several breaking points for information technology departments: the rise of "Shadow AI", which erodes traceability and cost control, model governance, still treated as an optional layer, and the AI Act, whose main risk will not be, according to the report, malicious non-compliance but accidental non-compliance, due to lack of preparation.
The report also highlights the Chief Data & AI Officer function, described as an arbitrator between speed and control, innovation and sovereignty, compliance and performance. As a member of the AI PACT, Sopra Steria indicates it has applied the regulation ahead of schedule to its own operations and is now supporting several major European accounts, particularly in banking and logistics, with their compliance programs.
