AI in the UK: a new mandate for trust

The recent UK AI Compass report gave one of the most detailed pictures yet of British attitudes to AI, based on a large, nationally representative survey that maps six distinct audience segments and tests what messages and policies resonate with each. It confirms a pattern seen in other UK surveys: most people recognise AI’s potential but are uneasy about its impacts on jobs, fairness and accountability, and they want more active regulation rather than a laissez‑faire approach.

This sits alongside government research showing that UK adults have mixed expectations about AI’s impact on society and on themselves, with many people unsure that current safeguards are adequate. For councils and care organisations, this is not an abstract policy debate: it is the backdrop against which they must introduce AI into contact centres, case management, assessment tools and back‑office automation.

As the UK AI Compass report shows, public consent is now the limiting factor for AI in UK public services, not technology. It shows a public that is engaged, segmented, and overwhelmingly expects stronger rules and accountability around AI, which should heavily influence how government and its partners design AI programmes. Datnexa’s view is simple: if local government and adult social care treat “public trust work” as a first‑class delivery stream, alongside data and engineering, AI can become an enabler of fairer, more human public services rather than a new source of risk.

Public services cannot “tech” their way out of a consent problem

The temptation in the public sector is to treat trust as something that will naturally follow successful pilots and improved outcomes. But the UK AI Compass work and related public attitude studies suggest the opposite: the starting point is scepticism, particularly where AI is invisible, involves vulnerable people, or touches decisions about support and entitlements. In adult social care or housing, communities already carrying the highest burden of inequality are also the most likely to be wary of opaque, data‑driven decision‑making.

AI governance initiatives and safety reports in the UK and internationally, underline that public legitimacy is now a core part of “safety”, not a communications afterthought. If local authorities deploy powerful models on top of fragile data foundations without visible safeguards, they risk not just legal or ethical failures but the political backlash that can freeze innovation for years.

What Datnexa thinks must change

Drawing on the UK AI Compass and parallel UK research on attitudes to AI, Datnexa believes public sector organisations need to reframe how they approach AI adoption. Three shifts stand out.

• Design with audience segments, not “the public”. The Compass identifies distinct mindsets towards AI, each moved by different stories, risks and benefits. Local government and ICS communications about AI should be designed with these segments in mind, for example, emphasising tangible safety and oversight for more cautious groups, and concrete service improvements for those who are curious but reluctant.

• Treat transparency as a product feature, not a compliance line. People say they are more comfortable with AI when there are clear laws, visible explanation and routes to challenge. For a resident using a council service, that means plain‑language notices about when AI is used, why, what data it relies on, and how a human can review or override an outcome baked into the service journey, not hidden in a privacy notice.

• Build consent and governance in the open. The Compass findings were shared with parliamentarians precisely because AI governance is now a public conversation, not just a technical one. Councils, combined authorities and ICSs should expect to publish AI registers, impact assessments and governance frameworks, and to co‑design them with residents, elected members, unions and voluntary sector partners.

  
  

A practical agenda for local government and adult social care

Trust‑centred AI needs to show up in delivery plans, not just strategies. Based on the emerging evidence, Datnexa sees five practical steps for public bodies.

• Start with AI where benefits are clear and risks are lower. Public attitudes data shows people are more positive about AI in areas like diagnosing illness or tackling fraud than in opaque, high‑stakes welfare decisions. Local leaders can sequence their AI roadmap accordingly, beginning with back‑office productivity or decision‑support tools under strong human oversight, before moving into more sensitive use cases.

• Build resident panels and lived‑experience groups into AI design. The UK AI Compass is a national‑level map - every place will have its own texture of trust, history and priorities. In social care and local government involving carers, disabled people, unpaid carers and community organisations early in AI pilots can surface concerns about bias, explainability and accessibility before they become operational issues.

• Make “explainable and contestable” a non‑negotiable requirement. Across multiple surveys, people say they want routes to challenge AI‑informed decisions and to have a human review the outcome. Procurement, architecture and assurance processes should encode that: no deployment of AI that affects access to services without clear lines of accountability, logging, and a tested mechanism for residents and staff to question and escalate.

• Align AI with data infrastructure, not work around it. Research on public sector data readiness for AI highlights widespread concern about patchy data quality, fragmented systems and limited governance. Trustworthy AI for social care and local services depends on the same foundations: robust shared data infrastructure, clear data ownership, and consistent standards across partners, not isolated proofs of concept bolted onto legacy systems.

• Equip staff to be honest brokers of AI, not just users. Frontline workers will carry much of the burden of explaining AI‑enabled tools to residents and families. Surveys show that trust often follows trusted intermediaries more than institutions or technologies themselves. Training, guidance and time for reflection are essential if staff are to use AI safely and confidently, and to push back when tools do not align with professional judgement.

  
  

From consent as a hurdle to consent as an asset

The UK AI Compass survey underscores that people’s attitudes to AI are not fixed; they shift when they see credible safeguards, tangible benefits and honest acknowledgement of risks. Public consent, handled well, can become an asset, a source of legitimacy for difficult changes, a compass for prioritising use cases, and a guardrail against technologies that look impressive but do not serve residents.

Datnexa’s position is that UK public bodies, especially in local government and adult social care, are now operating in a “trust‑first” environment for AI. Those who invest early in segment‑aware engagement, transparent governance and robust data infrastructure will be able to move faster, with fewer surprises, and with communities on their side. If you were to focus on one group in your area whose trust you most need to earn for AI in public services, who would it be?