AI in UK manufacturing: why governance decides who scales

UK manufacturers are not short of appetite for AI. They are short of the governance and operational foundations to scale it past a pilot. Make UK’s 2026 research found that just 2% of manufacturers have AI widely embedded across their operations, fewer than 40% use it in any area at all, and close to one in five have not adopted it in any form. Rockwell Automation’s 2026 State of Smart Manufacturing report, based on 1,560 manufacturers across 17 countries, tells a similar story from the other direction: 90% say digital transformation is now essential to staying competitive, yet only 34% of operations are AI-augmented today.

The gap between ambition and execution is not really a technology problem. It is a readiness problem, and it shows up in three places: where AI is actually being used, what is stopping it going further, and what happens when operational technology (OT) and IT are asked to work together for the first time.

Where AI already earns its place

Most of the value manufacturers report from AI so far sits in the back office, not the shop floor. Make UK’s data shows 83% of AI use is in functions like HR, finance and admin, against 11% in production, 7% in supply chain management and 6% in quality control. Rockwell’s numbers point the same way: 59% of manufacturers are actively using smart manufacturing technologies in some part of their operations, but only 18% remain stuck at the pilot stage, and the rest have not started.

The picture on the factory floor is changing, though, and not in the direction of headcount reduction. The World Economic Forum’s reporting on AI-augmented production lines describes a new kind of role emerging: process engineers who monitor multiple lines at once, spot subtle quality trends across batches, and flag problems between systems that were never designed to talk to each other. That is cross-system judgement, not data entry, and it is exactly the kind of work that is hard to automate and harder still to hire for.

The barrier is skills and governance, not appetite

Over half of the manufacturers Make UK surveyed cite skills shortages as the main reason they cannot move past small-scale trials. Nearly half expect AI to reshape jobs and working practices within two years, and most say they lack the time and clarity to build the data literacy, problem-solving and change management capability that scaling AI actually requires.

Rockwell’s findings add a second, related constraint: manufacturers are collecting more data than ever, but only 43% of it is used effectively, and poor execution explains that shortfall more than the data itself does. Nearly half (46%) had at least one cyber incident in the past year. Put those two figures together and the pattern is clear: manufacturers are asking AI to run on data and infrastructure that was not built with AI, or with the threats businesses face today, in mind.

OT and IT convergence is the risk most manufacturers haven’t secured

Manufacturing has a problem other sectors adopting AI do not share to the same degree. Much of its most important data, and increasingly its AI use cases, sit on operational technology: production lines, sensors and control systems built for decades of uptime, not for connecting to a corporate network. The UK government’s AI adoption plan for advanced manufacturing is blunt about this, describing manufacturing environments as “complex, highly engineered and often safety-critical,” and naming legacy operational systems and fragmented industrial data among the top barriers to safe AI rollout.

This is also where general AI security guidance stops being enough on its own. The NCSC’s guidance on AI and cyber security sets out risks specific to generative AI and large language models, including prompt injection, data poisoning and models that state incorrect information with total confidence. Its companion guidelines for secure AI system development, developed with international partners, and the government’s AI Cyber Security Code of Practice, published in January 2025, both push the same message: security has to be designed in from the start, and boards need to know where accountability for AI risk actually sits.

None of that guidance was written with a 20-year-old programmable logic controller in mind. For manufacturers, the more relevant reference point is IEC 62443, the international standard series for securing industrial automation and control systems across their full lifecycle. It treats OT as fundamentally different from IT, with its own performance requirements, equipment lifespans and consequences of compromise, and it is built around identifying high-value assets, assessing vulnerabilities and layering defences accordingly. Any manufacturer serious about scaling AI into production needs both frameworks: general AI governance for the data and models, and an OT-specific standard for the systems that AI is now touching.

What a safe path to scale looks like

The manufacturers making real progress are treating this as one problem, not two. They are getting an honest, evidenced view of where they stand against recognised frameworks, including ISO/IEC 27001, NIST and IEC 62443, while there is still time to close the gaps an assessment finds. They are building continuous compliance and evidence gathering into how the business runs, instead of scrambling for it once a year. And they are planning their cyber transformation roadmap alongside their AI roadmap, not as an afterthought once the AI project is already in production.

That is the work SysGroup does with manufacturing clients: a Cyber Maturity Assessment benchmarked against the standards that actually apply to converged OT/IT environments, followed by a transformation programme that closes the gaps it finds. It is less exciting than the AI use case itself. It is also the difference between a pilot that scales and one that stays a pilot forever.

If your AI roadmap is moving faster than your OT/IT security posture, that is worth finding out now, not during your next audit. Talk to SysGroup about a Cyber Maturity Assessment for your manufacturing environment.