Governments are under pressure to modernise. AI is moving quickly. Public expectations are changing with it.
But the challenge is no longer whether governments should use AI. It is how they can adopt it responsibly, confidently, and in ways that maintain public trust.
Trust Imperative 5.0, from the Boston Consulting Group, Salesforce, and the Centre for Public Impact, explores why many governments already have the right principles and governance frameworks in place, but still struggle to operationalise AI in practice.
The report examines how governments across ten countries are navigating questions around:
- risk classification
- decision-making authority
- human oversight
- accountability
- assurance for generative and agentic AI systems
It finds that the biggest barrier is often not technology, but operational clarity. When approval pathways, evidence requirements, and accountability are unclear, even lower-risk use cases can become stuck in duplication, delays, and uncertainty.
AI adoption needs to move at the speed of trust
The report argues that governments should avoid two extremes: moving so slowly that they miss opportunities to improve services, or moving so quickly that public trust erodes.
Instead, the focus should be on building “safe speed”: practical assurance models that help governments adopt AI responsibly while maintaining legitimacy and accountability.
That means:
- proportionate risk pathways
- clearer decision rights
- lifecycle oversight, not one-time approvals
- practical tools teams can actually use
- stronger human supervision for agentic systems
Why this matters now
Citizens are becoming more familiar with AI in their everyday lives and their expectations of government are rising alongside it. Trust increasingly depends on whether public institutions can keep pace while demonstrating that AI is being used responsibly.
At the same time, governments are facing growing pressure to improve productivity, strengthen service delivery, and respond to increasingly complex public challenges.
The next phase of AI governance is not about adding more policy layers. It is about making assurance usable.
Download the report
Trust Imperative 5.0: Building Trust in Government through Practical AI Assurance explores what is working in AI governance today, where friction remains, and how governments can adopt AI with greater confidence and public trust.
