Brits put more faith in AI than the government for facts, poll finds

Brits 'trust' AI over Government for 'correct facts' - what public servants should do now

A new study of 5,020 UK adults points to a shift: more people say they'd trust AI for "correct facts" than the Government. That's a signal, not a verdict - but it's loud enough to act on.

If you work in Government, this isn't about competing with chatbots. It's about rebuilding confidence in official information with proof, speed, and clarity.

Key numbers at a glance

• 44% believe AI will provide correct facts; 38% trust the Government.
• 57% think most public figures are intentionally selective with facts.
• 1 in 5 say AI is more trustworthy than friends and family.
• 70% have used AI; among users, 7 in 10 are confident in its answers.
• 25% trust facts seen on social platforms like X, Facebook, and Instagram.
• Biggest adopters: ages 25-34, with 66% placing trust in tools like Copilot, Gemini, and ChatGPT.
• Across all ages, 55% say it's getting harder to know which sources to trust.

As Harry Gove of OnePoll noted, "The abundance of accessible, multi-sourced information is reshaping how trust is formed and where authority is placed." That's the playing field now.

What this means for Government teams

• Assume AI is the first check. Citizens will verify your claims against an AI answer box. Publish information that's easy for people and machines to read, quote, and cross-reference.
• Trust now follows verification. Publish sources, methods, and timestamps with every major claim. Update visibly. Admit uncertainty early.
• Speed matters. Your "first version of the facts" should arrive quickly, then improve with follow-ups as evidence builds.
• Treat AI like a channel. Optimise pages so AI systems can extract unambiguous facts (clear headings, structured data, canonical pages).
• Meet the generational split. Younger adults are more ready to accept AI outputs and business leaders; older groups remain more sceptical. Tailor formats and channels accordingly.

Practical steps you can start this week

• Stand up "canonical fact" pages for high-interest topics with clear claims, citations, last-updated timestamps, and a changelog.
• Use structured data (clean headings, consistent definitions, downloadable tables) so facts are machine-readable and verifiable.
• Add content provenance signals to official media and PDFs to reduce spoofing and misattribution. See the open standard from the Coalition for Content Provenance and Authenticity C2PA.
• Create an internal AI playbook: approved tools, red lines, sourcing rules, and a human review checklist before anything goes public.
• Publish your method for statistics and estimates. Align with established assurance guidance where relevant, such as the UK's AI policy approach AI Regulation: a pro-innovation approach.
• Offer an "Ask us" route for complex questions and publish anonymised Q&A with sources. If AI assists, label it and note limitations.
• Upskill your team on prompt quality, verification, and safe use in public service contexts. Curated options by job role: Complete AI Training.

Guardrails for using AI in public service

• Human-in-the-loop. No unreviewed AI outputs in citizen-facing channels.
• Source policy. Rank official statistics, primary research, and signed statements above anonymous or unverified material.
• Evidence log. Keep prompt histories and citations for audit, FOI, and corrections.
• Data protection. Don't paste sensitive data into external tools. Use approved environments.
• Plain English first. Short sentences. Define terms. State what you know, what you don't, and what comes next.
• Clear labels. Mark estimates, modelled outputs, and confidence levels. Distinguish facts from commentary.

The generational gap to plan for

People aged 18-24 show higher trust in business leaders (49%) than those 55+ (39%). The 25-34 group is the most AI-forward. That means your channels, cadence, and proof need to work for both sceptics and digital natives - detailed briefs for some, crisp summaries and visuals for others.

Bottom line

Public trust is earned in the open. If citizens check AI first, make your facts easy to find, easy to verify, and hard to misinterpret. Lead with sources, speed, and transparency - and keep improving the proof as new data comes in.