The Public Accounts Committee (PAC) has urged the government to fix legacy IT issues if it wants to succeed with artificial intelligence (AI)-powered public sector initiatives.
The PAC’s Use of AI in government report noted that AI relies on high-quality data to learn. However, the committee was told by the Department for Science, Innovation and Technology (DSIT) that government data is often of poor quality and locked away in out-of-date legacy IT systems.
An estimated 28% of IT systems used by the government are legacy systems, covering both software and hardware. These, as defined by DSIT, include IT products that have reached end-of-life, are no longer supported by the supplier and are impossible to update. According to DSIT’s definition of legacy IT, these systems are no longer cost-effective or are considered to be above the acceptable risk threshold.
In its Digital and Data Roadmap 2022-25, the government committed to ensuring that all of the 72 highest-risk “red-rated” legacy systems across the government have remediation plans in place. However, the PAC reported that 21 legacy IT systems still lack remediation funding.
“DSIT told us that it was a matter of urgency that the issue of legacy systems in government is addressed, not only to take advantage of the opportunities offered by AI, but also to address other risks, including cyber security vulnerabilities,” the PAC report stated.
DSIT said it would prioritise “the systems that have the most valuable data” and “the highest levels of security vulnerability”.
From a legacy hardware perspective, the PAC said DSIT considers cloud services as a way for the government to move away from maintaining large datacentres. However, it noted that there is a trade-off between concentrating usage to achieve greater value for money and diversifying providers to seek greater resilience and security.
“As part of our inquiry into government cyber resilience, the Cabinet Office told us that, ‘quite a lot of our legacy systems are operating off a very small number of cloud providers, and until you fix some of the legacy issues, it is very difficult to move off some of the cloud services that are there’,” wrote the report authors.
According to DSIT, the new Government Digital Service (GDS) will work with HM Treasury to review all spending bids for legacy remediation to assess their credibility and ambition. The PAC noted that GDS will also be tracking the allocated money, which had too often been reallocated elsewhere in the past. There are also plans for pilots of funding on a continuous improvement basis to ensure that IT systems are maintained over the long term.
While the main issues in the list of barriers to public sector AI deployments concern legacy IT, the PAC warned that data sharing in government remains a problem.
A recently published blueprint for modern digital government sets out the government’s intention to work to reduce the barriers to sharing data by implementing standards, frameworks and tools. This includes the creation of a national data library to make it easier to find and reuse data across the public sector.
The PAC report recognises the importance of data transparency to improve trust. However, the report noted that as of January 2025, only 33 records had been published on the government website set up to provide greater transparency.
Along with the issues of legacy IT, data sharing and trust, the Public Accounts Committee also raised concerns about the digital skills gap. Despite government efforts to bolster digital skills in the public sector, the PAC said it remained sceptical that these reforms would address the issue where previous attempts have failed.
Among the recommendations made by the PAC, which include measuring the cost of legacy IT, the report said the government needs to establish a mechanism for systematically gathering and disseminating intelligence on pilots and their evaluation. The PAC also recommended the government identify common and scalable AI products and support their development and roll-out at scale.
Looking specifically at the procurement of AI, the report’s authors urged DSIT, in collaboration with the Cabinet Office, to set out publicly how its proposed AI sourcing and procurement framework will get the best from all suppliers in a market dominated by a small number of global technology companies and ensure opportunities are available for small suppliers.
