The Government Digital Service has updated its foundational guidance on digital technology governance, significantly expanding references to artificial intelligence. The update to the Government Functional Standard for Digital represents its biggest revision since coming into effect three and a half years ago, with AI references increasing from a single mention to more than 60 discrete references across the document.

The expanded guidance sets expectations for permanent secretaries, directors general, and all practitioners involved in planning, developing, delivering or managing digital, data and technology-related activity. The shift reflects a substantial increase in AI guidance across governance, data management, assurance, and service delivery sections.

From One Reference to Sixty: The Scale of Change

The Government Functional Standard for Digital, first piloted in 2020 and formally introduced in March 2023, previously contained only a single mention of artificial intelligence accompanied by a footnote. The latest iteration includes more than 60 discrete references to AI across the document.

The revised guidance addresses how AI will be considered in departmental strategies. According to the updated standard: "Each organisation should define and implement a government digital and data strategy, which shall… set out how AI will be used safely, lawfully, ethically and effectively, where relevant to organisational objectives."

In a note accompanying the changes, the GDS characterised the update as adding "proportionate references to artificial intelligence," indicating that the expansion reflects the scale of AI use across government functions without imposing uniform requirements across all bodies.

Governance Expectations Where AI Is Used

Where AI is used, the updated standard sets out expectations for governance. The guidance states: "Where AI is used, the plan should also define the intended benefits, data requirements, human oversight, assurance approach, model or tool ownership, monitoring arrangements and route for escalating risks. Plans should be agreed with relevant senior leaders and reviewed through governance and performance reporting."

The standard includes expectations for assurance: "Assurance should confirm that AI is used only where it is appropriate, lawful, secure and supported by the right skills and controls."

For service delivery using AI, the standard states that managers are advised to ensure "accuracy, reliability, bias, transparency and human oversight measures" are in place. Where analytics or AI are used in internal service monitoring, the guidance states that "outputs should be explainable, reliable and subject to appropriate human oversight."

Data Governance and AI Ethics Framework

The updated standard emphasises data governance as foundational to AI use. Organisations are directed to use the Data and AI Ethics Framework "throughout the planning, implementation and evaluation of projects, programmes, services, analytical work and AI-enabled solutions, organisations should use the Data and AI Ethics Framework to ensure data and AI are used appropriately, responsibly and transparently."

Data professionals are advised that "data assets used in AI systems are recorded with their purpose, provenance, quality, limitations, access conditions and retention requirements." The guidance also notes that "where data supports artificial intelligence tools, organisations should make sure it is lawful, relevant, accurate, secure and properly governed."

The standard acknowledges that AI monitoring can be used "where relevant, to prioritise iterative delivery and improvement" within service management. Where AI affects user experience and decisions, the standard emphasises that "AI should be used where there is clear evidence that it meets user needs and improves outcomes" and should be "supported by transparency where AI affects data, decisions, outputs or user experience."