UK Public Sector Must Build ‘Trustworthy, Open and Repeatable’ AI Foundations, Warns Red Hat

UK organisations have big ambitions for artificial intelligence (AI), but few are yet delivering real-world value, according to new research from Red Hat.

The findings reveal that 83 percent of UK respondents believe the nation has the potential to become a global AI powerhouse within three years. Yet almost 90 percent say their organisation is not yet delivering customer value from AI initiatives.

Jonny Williams, chief digital adviser for the UK public sector at Red Hat, said the results “mirror what we’re seeing across the UK public sector – organisations have big ambitions for AI, but are yet to turn investment into meaningful customer benefits. The question is how they can make the best use of resources to go from experimentation to effective business value at scale.”

Williams said the answer lies in focusing on three key principles: “We see the answer coming back to trust, efficiency and resilience. These are platform concepts that must apply from day one: choose flexibility, choice and transparency; set guardrails and governance; and upskill every team to use AI responsibly and effectively. This will help organisations replicate success, rather than duplicating efforts building multiple systems in silos.”

Skills Gaps and Shadow AI Persist

Skills remain one of the most pressing barriers, with 62 percent agreeing there is an urgent AI skills gap, particularly in agentic AI – systems that operate autonomously to perform complex, multi-step tasks – cited by 55 percent as the most in-demand capability.

Meanwhile, 83 percent of respondents report experiencing a “shadow AI” problem, referring to unauthorised use of AI tools by employees. Williams said this challenge should be seen as both a risk and an opportunity.

“While ‘shadow AI’ presents risks around data leaking and increased attack surface, it is also a sign that employees are eager to innovate, and leaders can use it to understand where there is a need for new functionality or for greater education around existing capability.”

Lessons from Cloud Adoption

Internal silos were identified in the survey as a top barrier to both cloud and AI adoption. Here, Williams warned that the public sector must avoid repeating past mistakes.
“We can learn from the government’s cloud adoption journey, where hundreds of departments developed their own cloud platforms, leading to a fragmented landscape that lacks interconnectivity and cannot easily take full advantage of the scale of cloud.”

He added that “AI presents a second chance to build more flexible foundations that can stand up to future unknowns, and start reusing common blueprints for success. Open source has a vital role to play here, enabling shared standards and common skills, helping make investment more consumable and reusable.”

Open Source at the Core

The survey found that 84 percent of respondents view enterprise open source software as important for AI strategy, with similar figures across hybrid cloud, virtualisation, and security.

Williams highlighted the potential for open collaboration in the public sector, pointing to initiatives such as the GOV Reuse Library: “A proposal to apply AI search capabilities to initiatives such as the GOV Reuse Library, for example, could deliver significant future benefit by making it easier for public sector teams to share components and accelerate delivery.”

He also stressed the need for flexibility and interoperability in AI models. “Organisations need to look at whether their AI is trained on the most valuable data, which is often going to be domain- and context-specific. That means they need to be able to choose the AI models that work for them and be able to switch between them. Again, open source and open standards can enable this interoperability.”