Somewhere in Singapore today, a citizen is using an AI assistant to find out what their retirement payout could look like. Instead of the official website or a certified factsheet, this interaction happens through a general-purpose AI tool, with immediate responses that help inform their decisions.
This is the permission paradox. Governments are investing heavily in AI infrastructure they govern and control, but citizens are increasingly receiving government information through AI systems that the public sector cannot train or vouch for with certainty. Today, AI platforms are increasingly influencing how information is discovered, interpreted, and trusted before a citizen reaches an official channel, if they reach one at all.
Singapore’s ambitions for AI-enabled government reflect a sustained commitment to digital capability building. Alongside over S$1 billion to its National AI Research and Development Plan through 2030, the upcoming Institute of Digital Government will equip more than 150,000 public officers with digital, data, and AI skills. But these efforts are directed inward at the infrastructure agencies control, while citizen behavior is shifting outward, toward platforms that agencies do not control.
As AI becomes more integrated into the citizen journey, the focus is extending beyond deployment toward accountability, orchestration, and trust. The next phase of digital government will depend on how effectively agencies connect data, content, and service delivery across increasingly autonomous and interconnected systems.
Better infrastructure, worse experiences
Adobe’s Digital Government Index (DGI) for Singapore, now in its third annual edition, reveals a public sector that continues to advance its digital maturity. In 2025, the overall score rose from 62.1 to 65.0, a 4.7 percent uplift driven by improvements in site performance and digital self-service capabilities. The trajectory signals sustained investment in the foundations of effective digital government.
The results also point to a more nuanced reality. In the same period, customer experience scores fell by 5.8 percent. The agencies that enhanced their technical infrastructure the most did not necessarily see corresponding improvements in how citizens experienced those services. In some cases, the experience became more fragmented despite stronger underlying infrastructure.
Across the assessed agencies, the gap between technical capability and experiential quality is consistent and widening. Many of the core components of digital government are largely in place, so the next phase depends on whether they are assembled in ways that serve the citizen, or in ways that merely serve the agency’s internal metrics.
When AI becomes the intermediary
Nearly four in 10 Singaporeans already use AI assistants instead of traditional search, while 18 percent are actively using agentic tools that can take actions on their behalf. Questions about tax obligations, healthcare subsidies, or education pathways, for instance, are now being interpreted through interfaces that sit outside the agency ecosystem.
AI intermediaries do not simply retrieve government information, they may reinterpret it. A general-purpose model trained on data from multiple sources, with no special obligation to current policy, will produce responses that feel authoritative even when they are incomplete, outdated, or contextually wrong. The citizen asking about retirement payouts may receive a confident, fluent answer that reflects last year’s parameters, or a misreading of eligibility criteria, which they are unlikely to know.
As citizens increasingly rely on AI intermediaries to navigate public services, the real challenge extends beyond visibility. It concerns accountability, including whether government information can retain accuracy and trust as citizen journeys become more distributed across external digital ecosystems.
Scaling AI with accountability
The launch of the Institute of Digital Government reflects this shift in emphasis. Requiring public officers to undergo mandatory training in cybersecurity, data protection, and AI literacy is a recognition that safe AI deployment depends as much on institutional judgment and operational readiness as it does on technology itself. But internal capability building, on its own, does not resolve the external dimension of the paradox.
The more pressing challenge is this: deployment cycles are accelerating faster than organizational adaptation. In this environment, where governments continue to accelerate their own digital ambitions, governance frameworks risk becoming retrospective, written to describe systems when they are already in operation. When that happens, accountability becomes difficult to locate and harder to enforce.
This is where orchestration becomes important. Agencies that are navigating AI adoption well are investing in stronger coordination across data, content, workflows, and service design. These foundations help ensure that AI-driven services remain accountable, coherent, and resilient as they scale. Over time, this becomes critical to sustaining public trust and maintaining confidence in the infrastructure.
Citizens are already ahead of the system
That citizen using an AI assistant to understand their retirement options is not an edge case. The example reflects a broader shift in how government information gets consumed, interpreted, and acted on. As AI becomes a more common interface for navigating public services, the question of who is responsible for what citizens are told, and what happens when they are told something wrong, is an immediate operational problem.
Singapore has the institutional will, sustained investment, and governance instincts to lead this transition in Southeast Asia. The DGI shows a public sector that continues to strengthen its digital maturity while maintaining visibility into areas that require attention. This provides an important basis for the next stage of transformation.
The challenge ahead centers on preparing government content and services for a world in which AI shapes the citizen journey. The permission paradox will not be resolved by building better agency platforms alone. It will require treating AI search surfaces with the same strategic seriousness as owned channels, designing for multiple platforms where citizens actually are. How governments respond to this shift will play a significant role in defining the future relationship between citizens and public services.
John Mackenney is Director, Digital Strategy Group – APAC at Adobe.
With over 15 years of experience in senior finance and digital roles, I am a passionate leader and advisor in customer experience transformation and digital innovation. As the Director of Digital Strategy Group APAC at Adobe, I leverage my credentials as a Chartered Global Management Accountant, a CPA, and a Graduate of the Australian Institute of Company Directors to support and guide the C-level executives of Adobe’s enterprise customers across various industries, especially in the financial sector.
My team and I use data-driven insights, digital benchmarking capabilities, customer journey mapping, and organizational assessments to deliver high-impact recommendations and solutions that enhance customer engagement, loyalty, and satisfaction. I draw on my practical implementation experience and knowledge of industry trends and best practices to help customers devise comprehensive and effective transformation programs that align with their goals and vision. Having led the digital transformation effort at Tourism Australia as the former GM of Digital Transformation and CFO, I have a first-hand understanding of how digital solutions can solve critical customer challenges and create better connected experiences for brands and governments across APAC.
TNGlobal INSIDER publishes contributions relevant to entrepreneurship and innovation. You may submit your own original or published contributions subject to editorial discretion.
Featured image: Alex Knight on Unsplash
What to do when the insider threat is no longer human, but a machine