
It’s a funny thing about AI but when I first started to blog about it the numbers were small.
But as time has gone on those numbers have grown slightly.
Talking to comms people there is weariness, nervousness and hesitancy.
I’m blogging this to light up the path ahead and for public sector people to understand how and why it can be used.
What the numbers say
I’ve gone through the numbers on AI-related surveys and insights in 2026 to try and cherry-pick what UK people think.
From this a head of comms, comms manager or comms officer can get a taste of where the public they serve is.
In short, the public are:
- Distrustful of AI and don’t trust the public sector.
- Want more regulation and guardrails around AI.
- Don’t mind individual AI uses.
- Want to be told when its being used.
- Don’t want to see it in content.
- Seeing deepfakes increasingly in local politics.
What this means for the public sector
When other recent innovations emerged, I’m not sure there was this level of distrust and pushback. The internet was largely welcomed. Social media grew an unstoppable momentum. But AI causes people to be nervous.
To nail the elephant in the room, this data is not an argument for using AI on the quiet without telling people.
Every piece of public sector guidance around AI stresses the importance of transparency. So comms must be transparent about how it uses AI itself.
New fields are emerging, not least how to influence AI tools that are searching for information. The gov.uk website is not the only source people draw from.
These stats are indicators for the path ahead.
UK survey data on AI as we head into 2027
There is a lack of optimism in AI
Stat: Daily users of AI are four times as likely to be optimistic about AI than those who have never used it. (Source: YouGov, 2026)
What this means: The figures play 37 per cent for daily users being optimistic against eight per cent for those who have never used it. That’s still not a majority for those who use AI extensively. So, if people in the organisation are advocates need to know that not everyone agrees with them. This should be an ongoing dialogue over what is being delivered.
Stat: One in 10 people say that deepfakes are a cause for optimism. (Source: Savanta for the ICO).
What this means: Deepfakes generate the lowest level of trust in technology. Don’t use these tools if you want people to trust you.
What people think about trust in organisations
Stat: Less than half of people trust the public sector with AI. (Source: NESTA, 2025).
What this means: Almost every stat in this round-up is from 2026. Why? Because the world is evolving very quickly. I’ve made the exception for this number published in December 2025. Why? Because it is a clear line in the sand. Almost half the UK population don’t trust the public sector with AI. That’s your starting point.
Stat: Nearly four times as many people distrust AI companies than trust them (source: YouGov, 2026)
What this means: If there is a strong distrust of AI companies then there is a distrust of AI. Public sector organisations need to build that trust. All the main public sector policies including UK Government, Welsh Government and Scottish Government stress openness when using AI.
We want more rules in place
Stat: Seven in 10 want stronger powers to regulators over AI content. (Source: Opinium, 2026)
What this means: The public want stronger controls in place over how AI is being used. While the direction for this is undoubtedly national there is absolutely a place for organisations to build trust by showing their own controls. We went to see those rules in place, please.
We don’t mind AI being used for specific things
Stat: Eight in 10 people want to see some element of AI in radiology for the detection of cancer. (Stat: Royal College of Radiographers, 2026).
What this means: If there is a clear purpose for public good then the UK public will support it. Usefully, this shows that attitudes to AI very much depend on what it is being used for.
Stat: Two thirds of people support AI in facial recognition by police. (Source: UK Government Home Office, 2026).
What this means: Again, people will be supportive of particular uses of AI rather than being blanket in favour. Just one in 10 were against this tactic.
Deepfakes have come to local politics
Stat: Almost a third of people saw a piece of political deepfake content in the run-up to the 2026 local elections (Stat: Demos, 2026)
What this means: Deepfakes have moved from national politics to local politics. Bearing in mind the lack of trust in deepfakes this can only have a negative impact on local politics furtrher eroding trust. Political parties and council standards boards need to consider how they regulate this area.
AI is becoming the news
Stat: One in five use AI to keep up with the news. (Source: Ofcom, 2026).
What this means: If a fifth of the population are using AI tools like Copilot, Claude or ChatGPT to understand the news then how to influence these search results are critically important.
Stat: Eight in 10 people in the UK are worried about AI making it difficult to know if the news is real. (Source: ONS, 2026)
What this means: If people are worried about AI content then the generation of AI content by public sector bodies in news is to be avoided. This would include images and video as two means of content that are easily ifdentified and untrusted. Words written with AI input would need to be checked and edited to ensure accuracy and that they meet house styles.
Stat: Nine in ten people want news organisations to mark news stories which have been generated by AI with no oversight. (Stat: YouGov, 2026)
What this means: People want to be told if AI has created a story. This stands at 91 per cent with no human oversight. This falls to 80 per cent if it is AI-written and checked by a human. Strongly, people want to be told when a human has artificial help. Public sector organisations using AI tools to create words should seriously think about creating content.
We’re not sure about AI content
Stat: Three quarters of the UK population don’t like the idea of AI actors in adverts. (Source: YouGov, 2026).
What this means: If people don’t like seeing AI characters in ads they are unlikely to like it in communications content. This is a powerful stat to use internally to people who think knocking out a quick AI pic of people is the answer. It isn’t.
We are using it more
Stat: Almost two thirds of people in the UK have used a generative AI tool. (Source: UK Government Department for Science, Innovation and Technology survey, 2026).
What this means: If 59 per cent of people have used such a tool then those who don’t have become a minority. This feels like a significant moment. If use is becoming widespread in the UK then the acceptance of generative AI tools – with guardrails – should follow.
For more, I deliver training to help you make sense of the changing landscape.
ESSENTIAL AI FOR PUBLIC SECTOR COMMS
ESSENTIAL COMMS SKILLS BOOSTER
ESSENTIAL VIDEO SKILLS REBOOTED.
Creative commons credit: London – 1983 by Helmut Zozmann, CC BY-SA 2.0 via Wikimedia Commons.
















