Innovation and regulatory evolution

Over the past year, the statistical system has continued to adapt through new methods, evolving practice and changes to the regulatory framework. This section considers three areas where that adaptation is most visible: the use of artificial intelligence, innovation and system capacity, and implementation of the revised Code of Practice for Statistics. It shows that the system is taking steps to respond to new opportunities and expectations, but progress is not consistent. AI use in official statistics remains limited, innovation is often shaped by capacity and prioritisation, and implementation of the revised Code is not yet fully in place.

Use of artificial intelligence

Artificial intelligence (AI) continues to attract interest across the statistical system. Current use in official statistics still appears limited, but activity is underway to understand where AI tools and models might be useful and how they should be applied. There is increasing activity across the Government Statistical Service (GSS) to understand current uses of AI, develop guidance and capability and to share learning across the statistical system. The Analysis Function is also developing wider guidance and resources on the safe and responsible use of AI. 

AI may offer opportunities to improve efficiency and support analytical work. For example, the Office for National Statistics (ONS)’s Survey Assist tool draws on ClassifAI to support the coding of survey responses. These types of tools may help reduce time spent on repetitive processing and allow statisticians to focus more on quality assurance and meeting user needs. At the same time, OSR’s current experience through regulatory work suggests that AI use in official statistics remains focused mainly on research and development, testing and administrative support rather than routine statistical production. 

Existing GSS leadership structures provide a route for understanding current uses of AI, and shared learning. The GSS has identified AI champions for 2026. The GSS and RSS Future Statistician vision also highlights the need for statisticians to embrace modern tools responsibly while protecting trustworthiness and integrity. 

However, there is still no single overall view of how much AI is currently being used, or could be used, across the statistical system. 

The Code of Practice for Statistics provides a robust and flexible framework that extends to the use of AI in official statistics. Where producers use AI in the production or communication of official statistics, they should apply the Code in the same way as they would for other statistical methods or models. To support this approach, OSR is updating its existing guidance for using statistical models so that it reflects the revised Code of Practice and applies clearly to AI tools and models. The updated guidance will continue to focus on Trustworthiness, Quality and Value, including the need for appropriate planning, transparency, quality assurance, human oversight and clear accountability for outputs. It will also support producers to consider whether use of AI is proportionate, how risks and environmental impacts are managed, and whether the approach provides clear value relative to its cost. 

AI is likely to remain an area of growing interest for the statistical system. Current evidence suggests a phase of coordination, capability building and experimentation rather than established practice. In this context, clear communication about how AI is being used, alongside consistent application of existing standards, will therefore remain important. 

Innovation and system capacity

Innovation is taking place across the statistical system, including in methods, data access and dissemination. In the examples we have seen, this activity is often closely tied to operational priorities and the practical constraints facing producers. Some producers are also using innovation proactively to support prioritisation and future resilience.  

There are examples of producers adapting their approaches in useful ways. The Department for Work and Pensions (DWP) has set out a programme of statistical work that includes greater use of reproducible analytical pipelines, linking survey and administrative data, and new approaches to user engagement. We have also seen examples of producers improving how users access and explore statistics, including in our assessment of Scotland’s 2022 census outputs, where we found that National Records Scotland had developed a range of innovative tools that enhanced users’ experience of using the data. 

Data sharing and linkage remain important to innovation and system capacity. Last year, we highlighted that progress in data sharing and linkage was often driven by individual successes rather than consistent system-wide improvement. There have been positive recent developments. The Department for Education has strengthened its long-standing collaboration with the UK Data Service on linked administrative datasets to support longitudinal analysis of education outcomes. ONS has also shared HMRC Pay As You Earn Real Time Information (PAYE RTI) administrative data with the Northern Ireland Statistics and Research Agency (NISRA) to support future population and migration statistics research and development. However, comprehensive and sustained system-wide data sharing and linkage still requires strategic leadership and buy-in from the wider data ecosystem, and cannot be achieved by statistics producers alone. 

There continues to be innovative practice across the system. A significant change this year was the introduction by ONS of groceries scanner data into consumer price inflation statistics in March 2026. This was a significant innovation in a core set of economic statistics, intended to strengthen the measurement of grocery price change through a larger and more detailed source of data.  

The statistical system continues to adapt, but the pace of innovation is shaped by available capacity and prioritisation decisions. We continue to consider that the recommendations from our 2024 follow-on report on data sharing and linkage remain relevant. 

Regulatory evolution

The revised Code of Practice for Statistics, published in October 2025, marked an important evolution for the statistical system. While retaining the core principles of Trustworthiness, Quality and Value, the revised Code provides greater clarity on how these principles must be applied in practice. This includes refreshed Standards for Official Statistics, which set out the practices required to ensure that statistics serve the public good and support public confidence. They also place greater emphasis on areas such as transparency, public involvement and engagement, and published statements of compliance. 

A key development in the revised Code is the introduction of the Standards for the Public Use of Statistics, Data and Wider Analysis. These standards set out what public bodies need to do to ensure an open, clear and accessible approach to the use of statistics, data and wider analysis in the public domain. They are based on the concept of intelligent transparency: proactively taking an open, clear and accessible approach to the use of data, statistics and wider analysis in public. In November 2025, OSR published the findings of its first annual review of intelligent transparency, which provided an initial picture of how far these expectations are being reflected in practice. The review made six recommendations aimed at helping intelligent transparency become the default approach across governments in the UK. 

There are signs that organisations are responding to these developments, and we note activity to support understanding of the new expectations beyond producer teams. For example, in response to OSR’s review of intelligent transparency, Number 10 appointed a chief analyst who is engaging with Heads of Profession for Statistics on the Standards for the Public Use of Statistics. These standards have also been acknowledged in ministerial responses to our report and related correspondence. 

Since its launch, all new OSR regulatory work has been based on Code 3.0, including assessments for accreditation and compliance reviews. In June 2026, OSR updated Heads of Profession for Statistics on its approach to applying Code 3.0 in regulatory work, building on the transition guidance published when the revised Code was launched. OSR has made clear that it is supporting the transition, recognising that embedding the new standards for existing statistics will take time.  

The revised Code requires producers to publish organisational policies and statements relating to statistics production. Producers must review current practices and publish required materials, and where required materials are not yet in place, producers should set out plans to address gaps with clear timeframes. For existing statistics, we expect producers to have clear, time-bound plans for implementing Code 3.0 where further work is needed. New and updated statistics should apply Code 3.0 from the start. OSR’s self-evaluation tool is one way of helping producers review how well their statistics meet the revised Code, both within teams and across the organisation, and identify where further development is needed. 

Initial regulatory application of the revised Code suggests that implementation is still at an early stage in some areas. The task now is to ensure the revised Code’s expectations are applied clearly and consistently in practice across the statistical system.