Tech outlook: What’s next for AI in government?

Governments are juggling rising public expectations, tight budgets, and legacy systems that were built decades ago and were never designed for today’s needs. Against this backdrop, AI can introduce enormous potential, but also significant risk. 

The stakes are high. For government especially, even a small failure can have outsized consequences, reinforcing public scepticism and resistance rather than support for innovation. Bias, opacity, and accountability missteps could quickly erode public trust, and once lost, this is extremely difficult to rebuild.

UK government is confident in its AI rollout – but may not be ready

The UK government is pushing forward with its well-documented plans to roll out an AI strategy. The Plan for Change states that AI will be used to deliver the UK’s national priorities, and position the country as an “AI maker rather than an AI taker”.

Its ambitions are wide ranging and significant, including plans to engineer new medical breakthroughs, and deploy tools that reduce emissions.

A March report from the Public Accounts Committee (PAC), however, is less confident about the government’s readiness for the strategy. It warned that an estimated 28% of government data was of poor quality in 2024, and often locked away in ‘legacy’ IT systems. The report notes that when it comes to the government’s adoption of AI, “the scale of the task ahead in grasping these opportunities is concerningly great.”

Is the government ready for AI in 2026?

Data is everything

The promise of AI in governments is immense, but it cannot be realised on a foundation of poor-quality data. 

To harness AI effectively, it’s crucial for governments to prioritise data hygiene. Their data must be consistent and accurate, and kept free from errors and biases. Without this foundation, even the most advanced AI systems will produce unreliable results.

Using raw PDFs for AI applications, for example, is a flawed strategy that introduces noise and discards critical context. Another step is necessary - purpose-built AI tools designed to structure and interpret complex documents can be key for processes like this.

Failing to build AI on accurate, clean data can lead to serious consequences, including biased decision-making, privacy violations, and operational failures. More than other organisations, governments can’t afford to risk experiencing these outcomes.

Back to basics

The true bottleneck for the government’s plans isn’t just the implementation of clever AI models. Instead, the first and biggest step is untangling decades of legacy systems and get humans and machines working in harmony.

Most government organisations still run on fragmented, ageing infrastructure built for an earlier era. Companies House, for example, was struggling with slow, manual processing of large numbers of physical documents. On average around 5,000 sets of accounts are filed electronically with Companies House each day, resulting in struggles with bottlenecks, delays in responding, and inefficiencies.

The key to improving this untenable situation was automating the processing of these documents. ABBYY’s AI-backed automation technology was drafted in to tackle the problem, and make all documents handled electronically. This greatly reduced the volume of paper moving around the building, whilst providing a cost-effective way to process forms and manage all types of complex business documents.

All of this modernisation applies to just one department – a lot of work still lies ahead.

Partnerships will be key

The UK government teaming up with Google Cloud to finally retire legacy tech sends a clear signal - in the public sector, modernisation and solid data foundations will matter far more than the sophistication of the AI models themselves.

The government says that its new partnership, announced in July, will help “modernise outdated government IT, upskill 100,000 civil servants in digital and AI by 2030, and secure better tech deals for taxpayers”.

Into next year, partnerships for cloud migration and workflow modernisation will accelerate, with companies like ABBYY positioned to bridge the gap between dusty legacy documents and modern agentic processes.

By transforming unstructured, outdated records into clean, structured, and usable data, these companies will enable governments to connect outdated processes with new, automated ways of working.

The smartest move government entities can make will be grounding AI adoption in clean data and process intelligence, building transparent models, and keeping humans firmly in the loop. Governments that get this right won’t just improve productivity - they’ll earn citizen trust, and maybe even start to make bureaucracy look fast.