Narrowing the UK defence’s AI capability gap

Embedding AI into UK defence systems can deliver measurable improvements to efficiency, accuracy, and decision-making quality once fully implemented. The UK Ministry of Defence (MOD) has made significant progress in establishing the policy, governance, accountability, and oversight architecture needed for AI adoption. 

Its Defence AI Strategy has laid out an ambitious position on personnel, data, procurement reform, and the cultural shifts needed to make AI a routine part of operations. That includes requiring each major MOD organisation to appoint a Responsible AI Senior Officer accountable for ethical compliance and escalation and to articulating its clear intention to embed AI into everyday decision-making and operations.  

While MOD organisations have made substantial progress in establishing AI governance frameworks and policies, they now need to put them into practice. To enhance military force effectiveness, they must incorporate AI into day-to-day workflows. 

Time to transition from policy to practice  

While MOD organisations have invested heavily in AI tools and platforms, they’ve seen uneven returns and inconsistent AI implementation. The structures needed to support faster progress and experimentation, including data infrastructure, workforce skills, and an AI-first organisational culture, are still in development. In early 2025, the House of Commons Defence Committee noted that the MOD had missed several digital strategy targets and concluded that cultural change would take longer than mandated. 

Since then, ministers have set clear, ambitious goals for force effectiveness and efficiency. But to achieve these, the MOD must deploy AI tools effectively and create operating conditions where they can be used routinely. At the moment, responsible AI governance, ethical principles, and senior-level ownership are in place, but the operational capacity is still being built. 

How to meet the challenges of transition 

To successfully make this transition, teams must have the data access, technical support, assurance pathways, commercial mechanisms, and incentives required to encourage safe, consistent AI use. Many public and private AI programmes have shown technical promise during early pilots only to lack sufficient attention to the conditions that promote routine use, such as data access, workflow redesign, and practitioner confidence.  

Organisations that focus on these conditions from the outset see faster, more durable adoption, while those that leave them for later often struggle to move beyond isolated pilots.  

Defence and national security organisations often face: 

• Insufficient workflow design, with AI pilots demonstrating promise but stalling when the underlying organisational workflows aren’t redesigned to reflect AI use 
• Fragmented data that’s often structured for storage instead of analysis, slowing efforts to generate reliable, operationally relevant datasets for use by AI tools 
• A shortage of trained personnel who can translate operational problems into AI-enabled workflows and support everyday AI users 
• Unclear ownership or misaligned incentivisation that can occur when teams that don’t directly benefit from AI use are asked to change their processes, causing hesitation in workplaces where assurance, procurement, and risk concerns already limit experimentation

What the UK can learn from the US experience 

We’ve seen AI adoption happen more quickly when organisations treat data readiness, workflow design, and practitioner capability as core delivery tasks instead of parallel workstreams. Progress accelerates when operating conditions are aligned with AI use at an early stage. 

The US defence and national security community has already navigated this transition to everyday AI use. For example, some US federal financial management programmes didn’t start routine use of AI tools until their teams had established predictable data pipelines, clarified ownership between mission and technical teams, and helped users apply the tools to existing decision cycles. This approach has often had more impact than the technical capabilities of an AI model design. 

Next steps for UK leaders 

For MOD investment in AI to produce measurable operational and efficiency returns, it must ensure that workflow processes, personnel, and data infrastructure are ready. UK leaders have already laid the groundwork by establishing sufficient guidance and governance. Now it’s time to help practitioners apply these structures in everyday settings. Progress will depend on: 

• Focusing on behaviour change management efforts rather than simply an accumulation of technology 
• Strengthening data infrastructure and architecture to give teams consistent inputs 
• Redesigning workflows so that AI contributes to decisions instead of running in parallel with existing processes 
• Clarifying AI ownership so that practitioners know who’s accountable for its use 
• Identifying user training needs and implementing initial and ongoing training programmes 

By focusing on how AI will be used and what its users will need rather than on the technology itself, defence organisations can translate intent into routine use, build practitioner confidence, and deliver measurable improvements.