New research commissioned by Red Hat has revealed that three-quarters of UK IT leaders are operating without strong artificial intelligence governance frameworks, raising concerns about organisational readiness, regulatory compliance and responsible AI deployment across the country.
The survey conducted by Censuswide across 500 IT leaders in the UK, France, Germany and Italy, highlights a growing disconnect between AI deployment and the level of organisational control. While 87 per cent of business IT leaders saying their organisations are already using agentic AI, only a quarter report having a strong governance framework in place.
Red Hat warns that many companies lack sufficient oversight of their data, infrastructure and relationships with AI providers, leaving them vulnerable as AI adoption accelerates.
Despite widespread concern about AI sovereignty and vendor control, the findings show that only 67 per cent of UK IT leaders have an AI exit strategy, meaning a third of organisations could struggle to switch providers without disruption.
For 43 per cent, such a switch would have a ‘moderate to significant’ business impact. Red Hat notes that this lack of preparedness underscores the growing urgency for organisations to strengthen governance and ensure they retain control over where their AI systems operate.
The survey also highlights that UK respondents are the most vocal in Europe in calling for stronger public policy on AI sovereignty. 89 per cent of British IT leaders want regulation that enforces open‑source principles, compared with 70 per cent in France and 72 per cent in Germany. Eight in ten decision‑makers believe companies should have greater control over building AI systems through open AI approaches, particularly as most major AI suppliers are based in the United States.
Stuart Harvey, CEO of Datactics, commented: “Many UK businesses don’t have clear visibility or control over their data and AI is simply exposing those weaknesses at scale, but the issue here isn’t just a lack of AI governance, but rather a lack of data readiness. If your data is inconsistent or poorly governed, deploying agentic AI will only amplify the problem.”
“The challenge is that AI governance is often treated as a compliance tick rather than something embedded into the data architecture itself. Static frameworks won’t keep up with fast-moving, autonomous systems. This is a systems and design issue and businesses need continuous oversight, strong data foundations, and flexible architectures, not just high-level governance plans.”
