The latest IBM Digital Sovereignty Index paints a stark picture for British enterprise. While eight out of ten senior managers claim they can keep artificial intelligence accountable, fewer than one in four can actually demonstrate that control through audits, regulator reviews or operational testing. The gap between confidence and evidence is widening as AI becomes the most fragile piece of the digital puzzle.
Confidence versus proof: the numbers that matter
According to the index, 84% of UK leaders say they are confident in their AI governance, yet only 22% can present verifiable audit trails that satisfy regulators or customers. Across the seven markets examined (the United Kingdom, United States, Germany, France, Saudi Arabia, India and Japan) a mere 32% of executives can provide strong evidence that their governance survives real-world pressure. The report defines digital sovereignty as an organisation’s ability to retain independent oversight, adapt systems and switch providers without losing operational control or breaching compliance.
AI registers the lowest control score
When the index grades four critical domains (data, operations, foundational technology and artificial intelligence) AI receives the weakest mark: 70 out of 100, the lowest of any category for the UK. Data, operations and core technology score 81, 80 and 79 respectively. The vulnerability is already surfacing: 93% of British respondents have encountered at least one sovereignty incident in the past three years, and AI-related problems were cited in 41% of those cases, edging out traditional cyber attacks (40%), provider changes (38%) and data-access disputes (32%).
How the United Kingdom compares globally
Despite the AI shortfall, the United Kingdom still ranks third overall, posting a composite score of 77 out of 100. This beats the global average of 74 and trails only India (79) and the United States (78). Moreover, 32% of UK firms meet the index’s “Leading” evidence standard, a benchmark that requires tested exit plans and traceable audit trails for AI-assisted decisions, compared with 26% worldwide. Stress-testing capabilities also show a modest lead: 39% of British respondents regularly simulate sovereignty scenarios, versus 33% of their international peers.
What the data says about organisational readiness
Only 26% of all surveyed enterprises reach the index’s Leading Digital Sovereignty tier, which measures fully integrated, demonstrable control. Companies in that tier are more than twice as likely to resolve their most severe sovereignty challenge without material disruption (52% versus 22%). Yet merely a third (33%) routinely test their ability to prove sovereignty, and 56% cite manual, time-consuming evidence collection as a major obstacle. Fragmented documentation, reliance on third-party records and legacy systems each hinder proof for roughly 53% of respondents.
Executive voices on the emerging risk
Leon Butler, CEO of IBM UK and Ireland, stresses that “real digital sovereignty means having the freedom to adapt, respond and change course without losing control.” He urges firms to embed openness, interoperability and flexibility from the design stage. Jonathan Adashek, IBM’s Senior Vice-President and Chief Global Affairs Officer, adds that sovereignty should expand future choices rather than restrict them, and that boards need concrete evidence of how they can switch providers or alter strategy under pressure. Emmanuel Frenehard, Chief Digital Officer at Sanofi, calls digital sovereignty “a business reality” and warns that without layered proof, AI disruptions can erode trust and continuity.
The IBM Index reveals a pronounced “overconfidence gap” for UK organisations: confidence is high, proof is low, and AI stands out as the weakest link. Enterprises that move beyond policy statements to tangible, regularly tested evidence will be better positioned to keep their digital operations resilient in an increasingly regulated AI landscape.
