- The UK MoD permits staff to use approved AI and LLM tools, subject to departmental, legal, and ethical requirements
- The department does not centrally record AI use at the level of individual contributions to published documents
- Human officials, rather than the AI systems, remain responsible for reviewing and verifying generated outputs
The UK Ministry of Defence (MoD) does not maintain any central recording or tracking of where, or how much, generative artificial intelligence (AI) and large language models (LLMs) directly contribute to published defence policy documents.
It was previously reported in 2024 by Army Technology that the MoD was utilising AI and LLMs to help sift through information for the subsequently published Strategic Defence Review.
However, new information gathered by Army Technology in a Freedom of Information (FoI) request sheds light on AI and LLM use by the MoD, where staff were “encouraged” to use such technology in areas such as planning, drafting, review, and editing.
According to the UK MoD, staff were permitted to use AI tools where their use was “authorised and compliant” with departmental policy, legal, and ethical requirements.
“Ministry of Defence staff have access to a range of approved AI tools for use in accordance with departmental policy,” the MoD said, following an FoI request regarding MoD use of AI.
Notably, the department does not centrally record or track the use of AI tools at the level of individual drafting contributions to specific publications.
Consequently, it is unclear to what extent documents such as the summer Defence Investment Plan (DIP) were drafted with AI assistance prior to final checks and publication.
In FoI responses specific to the DIP, the UK MoD said that it did not hold information identifying a specific AI or LLM used in its drafting.
It was however up to “officials” to check the accuracy of AI-generated content.
“Any outputs generated with the assistance of AI tools must be reviewed and verified by officials,” the MoD wrote in its responses to the FoI.
Further to its use of AI and LLMs, the UK MoD said the objective of its adoption was not “solely to achieve measurable cost savings on specific documents”, but also to “support productivity”.
Governance challenges mount
The use of AI by both private and public enterprises has accelerated in recent years, shaking up work processes and having a significant impact on job markets as functions deemed able to be performed by technologies like LLMs replace human functions.
According to recent analysis published in October by business intelligence firm GlobalData, calls for AI deceleration were “unlikely” to be heeded.
“Despite numerous AI scientists and developers voicing concerns over AI safety risks, the tech industry continues to pour hundreds of billions of dollars into infrastructure to develop increasingly powerful models,” GlobalData stated in its latest Artificial Intelligence Executive Briefing report.
