Public sector organizations worldwide are rushing to embrace artificial intelligence but lack the foundational data capabilities needed to succeed, according to comprehensive new research that surveyed 350 government agencies across six continents.

The Capgemini Research Institute study exposes a stark disconnect between governmental AI aspirations and execution capabilities. While nine in ten public sector organizations plan to explore agentic AI within the next 2-3 years, fewer than 25% report having the data maturity required to harness AI effectively.

"Governments everywhere are under immense pressure to deliver more – faster, and with fewer resources," said Marc Reinhardt, Public Sector Global Industry Leader at Capgemini.

"AI is the most powerful tool available in the arsenal, and data is an asset whose potential is underexploited today. But this transformation must be guided by strong AI governance, transparency, and a human-centered approach."

The research reveals that 64% of public sector organizations are already exploring or actively working on generative AI initiatives. Defence agencies lead adoption at 82%, followed by healthcare (75%) and security (70%). However, only 21% have progressed to pilots or deployment stages, and merely 6% have successfully put Gen AI into production.

The data readiness gap is particularly pronounced. Only 21% of surveyed organizations possess the required data to train and fine-tune AI models, while just 12% consider themselves mature in activating data for decision-making.

Australian Government Implementation

Several Australian government agencies demonstrating were cited for successful AI deployment. The Australian Taxation Office (ATO) has leveraged AI to detect $A530 million in unpaid taxes, halt $A2.5 billion in fraudulent claims, and achieve a 90% success rate in identifying superannuation underpayments.

The Australian Federal Police (AFP) has also embraced AI technology, using it to detect deepfakes and problematic content as part of their digital forensics capabilities.

"Data security, privacy, and timely data activation are all critical for public sector organizations," said Abhijit Gupta, Chief Technology Officer at Environment Protection Authority Victoria (EPA), Australia.

"A secure, modern, scalable, cloud-based infrastructure provides the appropriate foundations for developing this capability. Developing skills across the organization is vital, particularly for business users who need to interact with the data regularly."

Gupta emphasized the importance of specialized training: “This may include training in prompt engineering and other specialized skills to enable users to effectively access data and generate business value from its use. Finally, strong AI governance will ensure AI models are free from bias, risks have been considered, and data security and privacy are safeguarded."

Major Barriers to AI Implementation

The study identifies several critical obstacles preventing successful AI deployment:

Security and Trust Concerns: Data security issues top the list of barriers, cited by 79% of organizations, while 74% express limited trust in AI-generated outputs. These concerns stem from the need to protect sensitive citizen data and ensure AI system accuracy and fairness.

Data Sovereignty Issues: A significant 64% of organizations express concern about data sovereignty, with 58% worried about cloud sovereignty and 52% about AI sovereignty. This reflects governments' desire to maintain control over their digital infrastructure and data.

Budget and Infrastructure Limitations: Some 65% cite budget constraints as a significant barrier, while 77% point to the lack of modern, scalable infrastructure. Only 41% can access data at the speed required for decision-making.

Despite challenges, governments are investing heavily in data and AI leadership. The research shows that 64% of organizations already have a Chief Data Officer (CDO), with another 24% planning appointments. Similarly, 27% have appointed Chief AI Officers (CAIOs), while 41% plan to introduce this role.

"There is a strong focus on data and AI, especially with numerous central government announcements about transforming public sector services through AI," said Gurpreet Muctor, Chief Data and Technology Officer at Westminster City Council, UK.

"Excellent data management and governance are essential at both local and national government levels."

Regional Variations and EU Compliance Concerns

The study reveals significant regional differences in AI adoption. US government agencies are leading with 72% exploring or piloting Gen AI initiatives, compared to 55% in Europe. The Asia-Pacific region, which includes Australia, represents 14% of the total sample and shows strong practical implementation examples.

However, only 36% of EU-based organizations feel confident about complying with the EU AI Act, despite higher confidence levels for other data regulations.

National agencies outperform local ones, with 76% exploring Gen AI compared to 52% at the local level, suggesting budgetary constraints limit smaller agencies' AI adoption capabilities.

Data Sharing Challenges Persist

Cross-organizational data sharing remains problematic. While all surveyed organizations either have or plan data sharing initiatives, 65% are still in planning or pilot stages. Only 35% have rolled out or fully deployed data sharing programs, with just 8% achieving full deployment.

"It has become an undeniable truth that very few public sector actors have all the data they need to maximize their AI and data usage potential," said Peter Kraemer, Director of Data Sovereignty Solutions at Capgemini.

Looking Forward: Recommendations for Success

The study recommends a three-pronged approach for bridging the AI ambition-execution gap:

People-Centred Initiatives: Organizations should ensure clear vision and leadership, foster data-driven culture, and nurture analytical skills, especially among business users.

Process Reinvention: This includes implementing strong data governance with responsible AI practices and focusing on gradual data landscape modernization.

Technology Foundation: Investment in robust cloud-based data infrastructure and ensuring interoperability of data and IT systems.

"As AI begins to influence how decisions are made, policies are shaped, and services are delivered, governments have a rare opportunity to redefine trust, accountability, and citizen empowerment in the digital era," said Debarati Ganguly, Director of Data & AI for Global Public Sector at Capgemini.

The research was conducted between December 2024 and January 2025, surveying 700 executives from 350 public sector organizations across North America, Europe, APAC, and the Middle East, representing public administration, tax and customs, welfare, defense, security, and healthcare sectors.

Download the Report here.