SAS Global Study: Many Government Organisations Overrelying on Unproven AI

Government organisations demonstrate surprisingly strong overall Artificial Intelligence (AI) maturity, yet public sector investments in trustworthy AI technology and governance often lag behind. This suggests agencies may be deploying advanced AI built upon a shaky data foundation, increasing the risk of biased outcomes, security breaches, and costly operational failures. This is based on findings from a special public sector-focused report from SAS, Data and AI Impact Report: The Trust Imperative, with research insights by IDC.

The SAS report explores the “trust dilemma,” where organisations are either underusing reliable AI because they do not sufficiently trust it, or overrelying on AI systems that have not been adequately validated. This misalignment, evident across all regions, represents a critical barrier to effective government AI adoption. The new government report reveals trends in AI usage and investment across geographies, and how different regions are faring in closing the trust gap. While all industries in the report are struggling to overcome the trust dilemma, the public sector’s unique mission makes trust essential.

“For the public sector to rely on AI, it must deliver clear value while protecting the well-being of citizens,” said Grant Brooks, Senior Vice President of Public Sector at SAS. “Realising this value for citizens and communities requires aligning AI ambition and readiness. The report findings suggest we have work to do to achieve that.”

This special report by SAS contains regional analyses of North America, Europe, Latin America, META (Middle East, Turkey, and Africa), and Asia-Pacific. It includes survey results for each region’s public sector AI and data infrastructure, maturity, and trustworthy AI adoption, as well as more about the “trust dilemma.”

Balancing AI Speed with Public Accountability

Government organisations worldwide are rapidly adopting AI, with greater use of agentic AI (52%) than other major industries such as banking, healthcare, and retail. However, according to the report:

• Only 6% of government agencies are in the “ideal” state, with both high internal confidence in AI and AI systems that are demonstrably trustworthy, the lowest percentage of any of the industries included in the report.
• Thirty-eight percent of government organisations are both underutilising trustworthy AI safeguards and overrelying on AI, with a surprisingly high percentage placing strong confidence in AI systems that may not yet be fully trustworthy, such as GenAI.

In fact, public sector respondents trust GenAI far more than machine learning (ML). Despite years of proven ML use in functions such as tax and fraud detection, government leaders place more trust in a less explainable, more error-prone technology.

In terms of delivering trustworthy AI, government organisations lag behind insurance, banking, and life sciences. Only 15.3% operate at the highest level of the report’s Trustworthy AI Index, compared to the global average of 19.8%. They also fall short of banking and insurance organisations in their expectations for future investment in trustworthy AI initiatives.

“Government agencies are moving quickly from AI experimentation to operational use, but trust cannot be assumed, especially when systems influence public outcomes,” said Chris Marshall, Vice President, Data, Analytics, AI, Sustainability, and Industry Research at IDC. “Without strong data foundations and clear governance, confidence in AI can outpace trustworthiness, increasing risk for citizens and agencies alike.”

SAS Explores How to Close the Public Sector AI Trust Gap

While some governments are making progress in embedding trustworthy AI practices, most still face significant gaps in data centralisation, AI governance, and talent, which hinder their ability to fully realise AI’s potential. Government AI readiness and capabilities differ greatly by region, but there are some consistencies. According to the report:

• Every region cites a lack of a centralised or optimised data foundation as the top challenge to implementing AI.
• A lack of data governance is frequently cited as the second biggest challenge, with the exception of Latin America, where it is fourth.

Government organisations express strong expectations for AI investment growth in the coming year — 12.6% anticipate increases of more than 20%, and nearly half expect growth between 4% and 20%. Respondents view process efficiency and effectiveness as the most likely paths to realising AI’s business value. Additionally, personal productivity is cited by over 60%, the highest rate among the profiled industries.

In addition, government is the only sector where respondents are more likely to highlight skills gaps among general employee populations rather than among specialised technical teams. Reflecting these challenges, government organisations are prioritising investments in technology architecture alongside workforce skill development.

“Regardless of geography, many public sector organisations have bold plans to expand their use of AI in the coming years,” said Ravi Kant Sharma, Research Director, Government Insights, at IDC Asia Pacific. “The SAS report indicates they also understand their challenges. Investing in the right balance of infrastructure and trustworthy technologies will be critical to successful AI deployments.”