Public consent for AI in the UK is not yet won, not yet lost, and not yet being shaped by the people building the technology. New Teneo research — combining a nationally representative survey, an MP poll, a sample of UK technology professionals and a randomised control trial of pro-AI messaging — finds that the British debate sits on a knife edge, that the tech sector is misreading what worries voters and ministers, and that only one argument tested actually shifts opinion. For UK AI businesses, this is the difference between earning a public licence to operate and watching policy harden around a constituency they never spoke to.
The communications problem hiding in plain sight
UK AI businesses talk about public consent the way they talk about cybersecurity in 2019: a thing that matters in principle, addressed by press releases and the occasional white paper, owned by nobody in particular. Teneo’s research suggests that approach is now actively dangerous. Britain’s swing voters on AI — the persuadable middle who will decide whether regulation tightens, loosens or fragments — are demographically and economically distant from the sector arguing for them. They do not work in technology, do not benefit obviously from it, and do not respond to the arguments most senior tech leaders believe will work.
Strategic Reality: Public opinion on AI in the UK is finely balanced rather than settled. That makes it contingent on communication, not inevitable on capability. Capability advances will not, on their own, deliver consent.
The Teneo work matters because of its method. A randomised control trial design measures actual attitude shift from exposure to specific arguments, not stated preference. A 2,004-person nationally representative sample, 102 MPs and 502 technology professionals lets the research triangulate where each group sits and where they diverge. The headline finding is uncomfortable for anyone in or around UK AI: the people building the technology are the least accurate predictors of what persuades the people who will decide its fate.
The real story: jobs is not the worry, safety is
The dominant boardroom assumption about AI’s public-perception problem is that jobs sit at the top of the anxiety stack. The Teneo data does not support that. Job displacement features in the picture, but it is not the most pressing worry for either MPs or the public — including among people who use AI in their daily lives. The active worries are safety, misuse, fraud and loss of control.
This matters because messaging built on a jobs frame (“AI will create new roles, reskilling, transition support”) is answering a question that is no longer top of mind. It also misallocates communications spend: AI businesses produce thoughtful workforce transition content for an audience whose actual unease is about being defrauded by a synthesised voice, about systems failing in ways nobody can explain, about decisions being made by machines that cannot be held to account. The two conversations look adjacent but are not interchangeable.
Where UK opinion sits today
| Group | Primary AI concerns | Where messaging often lands |
|---|---|---|
| UK public (2,004 surveyed) | Safety, misuse, fraud, loss of control | Productivity gains, jobs upside |
| UK MPs (102 surveyed) | Safety, misuse, public-service integrity | Innovation, investment headlines |
| UK tech professionals (502 surveyed) | Capability, competitiveness, regulation drag | Self-evident benefits |
The pattern is consistent: the audiences AI businesses need to persuade are worried about a different category of problem than the one most pro-AI arguments are built to answer.
Deep dive: only one argument changed minds
The Teneo team tested a series of pro-AI arguments under a randomised control trial design. Only one produced a statistically meaningful shift in public opinion. That argument moved overall support for AI from a “referendum-losing” 45% to a “referendum-winning” 56% — an 11-point swing that held across party affiliation and demographic group, and which the same research found to be the most persuasive argument among MPs as well.
The winning argument was AI’s potential to deliver tangible improvements in public services — and specifically the NHS. Not abstract transformation. Not GDP. Not productivity. Shorter waiting times. Greater efficiency. Better frontline delivery.
Strategic Insight: A relatively modest, concrete promise about a service the public uses and cares about beat every more dramatic vision tested. The decisive frame for UK AI is operational, public, and verifiable — not transformational, private, or speculative.
That is a counterintuitive result for an industry whose communications instincts run in the opposite direction. UK AI businesses tend to lead with productivity multipliers, economic upside and “transformation” — abstractions that test well in a boardroom but, by this evidence, do not shift voter or MP opinion. The NHS frame works because it is testable, locally felt and ethically defensible. People know what a shorter wait looks like.
Why the NHS frame holds across the political map
The 11-point swing held across partisan divides. That is unusual. Most pro-AI framing tests well with the demographic groups most receptive to AI in the first place (younger, more educated, urban, economically secure), and falls flat with everyone else. The NHS frame closed that gap. The implication for UK AI businesses is that public-service improvement is the rare argument that does not require a sender to choose a political lane.
Strategic analysis: the tech sector’s blind spot
Among the Teneo research’s most striking findings: fewer than 2% of UK technology professionals could correctly identify more than half of the arguments most likely to persuade either the public or politicians. And the picture got worse with seniority — the more senior the tech professional, the less accurately they predicted what would move opinion outside their own sector.
This is not a knowledge gap. It is an audience gap. The technology sector’s social, professional and informational networks reinforce a set of pro-AI arguments that are coherent internally and unpersuasive externally. Senior leaders, who spend the most time in industry conferences, board rooms and peer conversations, end up the most insulated. They are confidently advocating for arguments that the research suggests do not work, and they do not know it.
| Stakeholder group | What they believe will persuade | What actually persuades |
|---|---|---|
| Tech founders / CEOs | Economic competitiveness, sovereignty | Public-service outcomes |
| Tech investors | Productivity multipliers, market size | Operational reliability stories |
| Policy / public affairs leads | Innovation framing, jobs transition | NHS / public services use cases |
| External-facing comms teams | Aspirational transformation language | Concrete, verifiable improvements |
The structural problem here is that the people setting communications strategy in AI businesses are typically the people the research finds to be the worst predictors of public response. Without a deliberate corrective (independent research, audience testing, lay-audience listening), the gap will not close on its own.
Hidden Cost: Every communications cycle that runs on tech-sector assumptions burns political capital. Each one fails to move the swing voters whose views will harden over time, making future persuasion harder and regulation more reactive.
Strategic recommendations: building a UK AI narrative that actually works
The Teneo research does not prescribe a comms plan, but the implications for UK AI businesses are direct. Treat the following as a working framework rather than a finished playbook.
Priority actions for AI businesses
For commercial AI companies operating in the UK:
- Audit your external-facing messaging against the safety / misuse / fraud / loss-of-control axis. If your top-line copy is built around productivity or jobs, you are answering a question voters are not asking.
- Identify the public-service or public-good application of your technology (even if it is secondary to your commercial use case) and develop genuine, verifiable case studies around it. Synthetic NHS framing without underlying work will be detected and punished.
- Test pro-AI messaging on a lay UK audience, not on internal stakeholders or sector peers. The research suggests internal validation is actively misleading.
For UK AI policy and government affairs functions:
- Reframe parliamentary engagement around public-service delivery rather than competitiveness. The Teneo MP data suggests this is the frame that resonates across party lines.
- Bring operational evidence (not just economic forecasts) to ministerial conversations. MPs respond to the same things voters do: tangible, locally felt improvement.
- Build relationships with public-service-adjacent stakeholders (NHS trusts, local government, public-sector unions) before you need them. Consent is built before crisis, not during it.
Priority actions by organisational maturity
| Maturity level | First move | Second move |
|---|---|---|
| Early-stage AI business | Find one genuine public-good application and document it | Test messaging on non-tech audiences before launch |
| Scale-up | Audit existing comms against the safety/misuse axis | Resource an external-affairs function with non-tech background |
| Established AI company | Independent audience research, not internal review | Invest in public-service case studies as a strategic asset |
| Public-sector AI vendor | Lead with operational reliability and accountability evidence | Treat NHS-style framing as core, not adjunct |
Success Factor: The single most useful investment a UK AI business can make this year is independent audience research with non-technology audiences. Not focus groups dressed up for the board, but research designed to find out what voters and MPs actually respond to, and which of your assumptions are wrong.
Hidden challenges most strategies miss
The Teneo findings open four less-discussed problems that UK AI businesses will need to address if they want to operate in a benign policy environment.
1. The asymmetry of negative events. A single high-profile fraud, safety failure or public-service deployment gone wrong will shift opinion faster and further than any positive case study. The 11-point swing the research measured can be reversed by one bad week. Sector-wide reputation depends on the weakest deployment, not the average.
2. The credibility tax on tech voices. The same research that finds public-service framing persuasive also implies that tech-sector messengers are not the most credible carriers of that message. NHS trusts, frontline clinicians, local government leaders and independent researchers will move opinion further than tech CEOs saying the same thing. Building those messenger relationships is slow, multi-year work.
3. The MP–voter alignment risk. Teneo’s data shows MPs and the public broadly agree on what worries them and what persuades them. That alignment is unusual and load-bearing. If a populist political incentive emerges to break it (for example, an opposition party finding electoral advantage in a hard AI-safety position), current pro-AI consensus could fragment quickly.
4. The senior-leader problem. Because the research finds senior tech professionals to be the least accurate predictors of public response, the people setting communications strategy are structurally the wrong people to validate it. Organisations need a governance mechanism that gives external research and non-tech voices actual veto power over messaging, not just an advisory role.
Reality Check: Most UK AI businesses currently have communications strategies signed off by exactly the cohort the Teneo research identifies as least able to predict what works. That is a fixable problem, but only if leadership accepts it is a problem.
Strategic takeaway: persuasion is a capability, not a campaign
The Teneo research’s most important argument is one it makes by implication rather than statement: persuasion, not innovation, will determine whether AI businesses operate in a benign or hostile UK environment over the next five years. Capability gains do not, on this evidence, convert to public consent. Communications has to do that work, and the work is harder than the sector currently assumes.
Three success factors for UK AI narrative-shaping
- Lead with what voters actually care about. Safety, misuse, fraud and loss of control are the active worries. Address them directly in messaging rather than routing around them with productivity arguments.
- Choose the NHS / public-services frame deliberately. The data on this is unusually clean: it works, across party lines, in a way nothing else tested does. Treat it as core messaging architecture, not a sector campaign.
- Get the strategy out of the boardroom. Senior tech voices are the least reliable validators of pro-AI messaging. Independent research and non-tech audience testing are now operational requirements.
Next steps checklist for UK AI leaders
- Map current external messaging against the safety / misuse / fraud / control axis
- Identify and document a genuine public-service application of the technology
- Commission independent audience research with non-technology audiences
- Build relationships with public-sector and frontline messengers before they are needed
- Establish a governance mechanism that gives external evidence veto power over comms strategy
- Stress-test the organisation against a single-event reputation shock scenario
Take Action: The Teneo research has done a meaningful chunk of the audience work for the UK AI sector. The competitive advantage now sits with the organisations that act on it before their peers do, and with those willing to be specifically wrong about their existing assumptions.
Source citation & attribution
This analysis draws on research published by Teneo, Persuasion with Precision: Winning the AI Argument in the UK (11 May 2026). Teneo’s methodology combined a nationally representative survey of 2,004 UK adults, a poll of 102 Members of Parliament, a dedicated survey of 502 UK technology professionals, and in-depth qualitative interviews with AI communications specialists, with pro-AI argument testing under a randomised control trial design.
The strategic analysis, business implications, frameworks and recommendations are original work by Resultsense, interpreting Teneo’s research findings for UK AI business leaders, policy professionals and communications functions. Resultsense is a UK-focused AI news and analysis publication. See our Insights coverage for further strategic perspectives on UK AI policy, governance and adoption.