How to medicate those sclerotic, wheezing twins, the British economy and state? This has been the central question facing successive UK governments since at least the financial crisis of 2008 – and the question to which no prime minister has been able to provide a convincing answer. At least, that is, until November 2022 and the coming of age of Silicon Valley’s latest elixir: artificial intelligence. Here, suddenly, was the wonder pill that could transform everything. Rishi Sunak was the first prime minister to embrace its transformative qualities. Tony Blair declared confidently that soon little else would matter. And then, finally, Keir Starmer threw his weight behind it. The British state had found its cure-all.
If the AI roll-out amounts to a social and economic revolution, then we must prepare for the political reaction bound to follow, but which few in the political world understand. The “techlash” has already begun, and it may yet prove as transformative for world politics as the technology itself claims to be.
Put your ear to the ground and under all the noisy enthusiasm a shift is already audible. In March last year, the Ada Lovelace Institute – which is dedicated to ensuring that AI works for those of us who don’t run tech companies – found that concern around how AI would be used had increased “significantly” since its first study in 2022-23, when ChatGPT was launched. In a further study last December, the same institute found that 84 per cent of the public believed the government put the interests of Big Tech ahead of their own. And that was before January’s torrent of sexualised deepfakes, generated from images of real people with the help of Elon Musk’s AI service Grok, and the growing controversy around Palantir, the data analysis firm that works with Ice in the US, the Israeli defence ministry and the NHS. According to one parliamentary aide, public disquiet about both companies’ behaviour has been landing in MPs’ in-trays.
The public’s objections to the AI revolution are widespread and growing, from the copyright violations involved in training AI models – in which giant tech firms trawl through books, music, newspaper articles and so on to feed content into their large language models – to the environmental impact of data centres, and the semi-autonomous weapons systems being used on the front lines in Ukraine and in Gaza. AI slop pervades social media and, increasingly, every other form of media, from film to music, journalism and even books. There is also mounting concern about the racial bias in AI facial recognition technology, the dangers of driverless vehicles and the potential destruction of vast swathes of productive work. Beyond all that lurks the most frightening spectre of all: the prospect of handing extortionists, terrorists and dictators the means to hack into our critical infrastructure; to control military hardware and create bioweapons; even to launch wars. Oh, and of course, the threat of turning humanity into one great Dr Frankenstein, as we lose control of our superintelligent creation altogether.
In the US, the techlash is already clearly evident. In April, OpenAI CEO Sam Altman’s home was firebombed. An independent report by the research group Data Center Opposition claimed membership of groups opposing data centres had quadrupled in six months to more than 360,000, and voters in Missouri threw out city councillors for ignoring such objections. In July, New York became the first state to impose a moratorium on the construction of large new centres. Think tanks suggest the UK may be taking a similar turn. In a report for Policy Exchange, the tech expert and former senior Conservative adviser Ed de Minckwitz identified a “significant political movement” emerging across the ideological spectrum, which didn’t necessarily oppose technological advances, but was fighting “for the right to opt out of AI-mediated existence”. A paper by the Institute for Public Policy Research (IPPR), “Acceleration Is Not a Strategy”, warned that if the government fails to put the public interest first and simply lets AI tear through the economy, “public pressure for blunt, populist measures, from data-centre moratoriums to outright bans on AI deployment, is likely to grow”.
Techno-optimists might have hoped that young people would be enthusiastic about AI – and many are. Much to the optimists’ surprise, however, there is also vehement pushback from a generation that has already had other problematic technologies imposed on it. The independent tech and AI researcher Renate Samson worked with the Nuffield Foundation to ascertain the attitudes of 14- to 24-year-olds to digital life, and found a widespread sense of disempowerment. “In many conversations,” she says, “it eventually comes round to: ‘It’s capitalism isn’t it? Burn it all down. Turn it all off.’”
For those already well into their careers or approaching retirement, comfortable in their homes, AI may be little more than a clever Google: a tool to plan holidays, ask advice and, perhaps, for some, find companionship. But for a generation who have grown up with technology invading every corner of their lives and are heavily indebted from university, unable to find work or afford to buy their own homes, the latest technological revolution looks less like an exciting new toy and far more like another threat to their agency, quality of life and the society they want to live in.
Mainstream politics has not yet responded effectively to this ferment of popular disquiet. Zack Polanski has been vocal in his attacks on Palantir, and has criticised the environmental impact of data centres and the threat AI poses to creatives. But the subject has not been central to the Green Party’s messaging. There have been signs of Reform considering the populist potential of this agenda, but that appears to be in tension with its dreams of using tech to make the state more efficient. On Labour’s increasingly influential soft left, there is revulsion at the Grok row, and concern about environmental impacts and overreliance on companies rooted in an ever more wild America, but no agreed position.
Yet as AI’s capabilities soar above even recent expectations, its political salience is unlikely to fall – certainly not before the next election. And elsewhere, there are signs that disquiet over the risks is beginning to cohere.
Joseph Miller, now a doctoral student at Oxford University, founded an AI-driven text-to-image website before moving into AI safety research. In 2023 he became the founding director of PauseAI UK. This is one of several campaign groups led by young people with backgrounds in AI that have appeared in the past three years. Each is motivated by the same basic shock: that the technology is developing far faster than expected, endangering our ability to keep control over the systems we all rely on. They worry that, unless we act soon, it will be too late to put safeguards in place. And they charge that promises to tread cautiously by CEOs such as OpenAI’s Altman and Google DeepMind’s Demis Hassabis have been forgotten.
PauseAI UK’s first protest, outside DeepMind’s London office, drew all of five people. But over the past year, it has organised more, helped to stage a debate on AI safety in Westminster Hall and coaxed 60 British politicians to sign an open letter to Google DeepMind demanding greater transparency. The leaders of DeepMind and Anthropic, Miller told me, have “explicitly said, ‘We would like to go slower, but we can’t, because we’re competing with China,’ and that’s a crazy situation.” (Since Miller and I spoke, Hassabis has called for international consensus “on how to manage the most serious risks”.)
Andrea Miotti worked in AI safety before founding his organisation, ControlAI, in October 2023, ahead of the first ever AI Safety Summit at Bletchley Park. Miotti wanted to make sure the international gathering of government officials, brought together by the Sunak government, discussed the “risk of extinction” from superintelligence. He sees AI’s strengths, particularly its narrower applications, but argues that autonomous “systems that are vastly smarter than humans” could quickly leave us “sidelined from decision-making”, including on military and economic issues. He fears humans might even lose control over AI systems we can’t understand, to the point that we end up “living like pets”.ControlAI’s newsletter goes out to around 45,000 people in the UK, and three times that in the US. Its work has prompted debates in the House of Lords; more than 100 parliamentarians have signed its call for “binding regulations on the most powerful AI systems”.
The case has only been strengthened by the recent disturbing revelations that – in a grim global first – AI models created by several of the dominant tech companies have gone rogue. True, the AIs were following human instructions – but they did so by finding supposedly impossible ways to break free of safety constraints with dizzying ease, before swiftly hacking other companies’ systems.
Until recently, campaigners’ focus has been split between existential risk and more immediate harms. But Elizabeth Minor, who is head of policy for the international umbrella group Stop Killer Robots, says that as public awareness rises, a significant shift is taking place. Just as the behaviour of tech companies and governments shows “that all this stuff is quite connected”, the various campaigns are becoming more connected, too.
This year, a new group emerged that embodies this. Pull the Plug was set up by Harry Atkinson and Frieda Lürken. They do not wish to pull the plug on AI itself, but on the concentrated power of its CEOs. Atkinson was driven to organise by the impact of compulsory AI use on his work in copywriting, and by friends’ stories from the film industry. Lürken is a co-founder of Extinction Rebellion, and objects to the building of hugely expensive, energy-intensive data centres on the basis of speculative demand, in the face of local opposition. (With around 500 sites, the UK is reportedly the third largest data-centre hub in the world.) Lürken does not, at this stage, see a need for civil disobedience, but, like Extinction Rebellion, Pull the Plug calls for citizens’ assemblies – to give people a say.
In February, they worked with PauseAI to organise the largest British march so far, of several hundred people, past the London headquarters of OpenAI, Meta and DeepMind. Atkinson argues that the range of causes that motivated people to turn up illustrated the “broad church” now congregating around the issue. As Lürken told me: “Any successful big movement against huge power needs a big coming together of groups who don’t agree on everything.”
Denys Sheremet is another campaigner with an AI background who was driven to protest by his fear of AI’s existential dangers. He spent a month on hunger strike outside Google DeepMind. His group, Together Against AI, has concluded that the public is “way more anti-AI than we originally thought”.
For the moment, these groups remain on the fringe. But they have already made swift advances, and as AI’s revolutionary effects ripple across the economy and society, the techlash could soon expand. James Kanagasooriam, chief research officer at Focaldata – and the man who first identified the “Red Wall” – predicts the way people feel about AI will soon become an “axis in our politics”. If we’re producing a generation of graduates who will simply never enter the professional elite because AI has left it so shrunken, he says, that “could be transformative”.
Even while Starmer was still in office, there were signs that the backlash was beginning to register in Westminster. Among the most high-profile supporters of AI in government, boosterish rhetoric was giving way to acknowledgement of public unease. By its final months, the Starmer government appeared to have accepted that simply lecturing the public on their need to get on board had backfired. In one of her last speeches as technology secretary, Liz Kendall sympathised with public concern, but also condemned calls to “pause AI” as a “betrayal of British talent and British interests”.
Partly in response to anxiety about Britain’s reliance on a now unreliable America, Kendall argued that Britain should assert its sovereign control over its AI ecosystem. The problem is that, beyond small specialist companies, we really have no such thing. Since the coalition government allowed Google to swallow DeepMind in 2014, Britain has not come close to creating a broad-ranging AI foundation model. France’s Mistral may be more modest than the US and Chinese leviathans, but it is significantly more than we have. In December, Focaldata found that while digital sovereignty was an issue new to most voters, strikingly high proportions wanted more safeguards. In May, Antonio Weiss, the author of AI Demystified who used to advise Labour on tech, told me that there are points in the value chain where Britain could plausibly reduce its dependencies. Some argue that our dependence on US systems is now so deeply entrenched that it is simply too late. But whether the technology is British or foreign in origin, the danger is that, because many people already feel disempowered, the advance of AI will feel as if it is just punching their bruises.
It is not yet clear how the new government will approach AI. Andy Burnham’s emphasis on growth that benefits the population, and his move to scrap the Department of Science, Innovation and Technology, suggest a more sceptical approach. He has, however, appointed a minister for AI who will attend cabinet, the respected Kanishka Narayan.
Meanwhile, the stakes are growing ever higher. AI’s cheerleaders proclaim that its repercussions will match or exceed those of the Industrial Revolution. In 2015 Altman was already worrying that “the software revolution” would “create wealth but destroy jobs”, concentrate wealth and trigger “drastic wealth inequality”. Blair made the same analogy in 2023, noting that, as in the 19th century, the revolution would “change the state”. This year, Dario Amodei of Anthropic and Elon Musk have invoked the ructions triggered by the spinning jenny, the steam engine and the blast furnace – and hundreds of experts have used the same comparison to call for urgent preparation for AI’s potential devastation of the labour market.
But there is rather less talk of the political upheavals that the first Industrial Revolution provoked. Blair has noted that it led to the creation of the Labour Party. But of course it triggered a great deal more than that – not least the anger of the Luddite weavers, who followed the legendary Ned Ludd, in smashing the machines that were taking their jobs. It also contributed to the birth of the modern city, the railways, even mass democracy.
This history suggests that there are new seismic ruptures coming our way – though they may not be quite so constructive. Whereas television took off in the placid 1950s, and the internet landed in the optimistic 1990s, AI has crashed in on a world already on fire. The climate emergency is growing ever more threatening. Trust in the power of democratic politics is badly weakened, and political violence is rising. Respect for science retreats before waves of conspiracy theory. And now graduates, already struggling with student loans and an impossible housing market, are watching the entry points to the elite career paths they thought they could rely on vanish.
Historically, as the scientist Peter Turchin has charted, “elite overproduction” has destabilised societies. A 2026 study by King’s College London’s Policy Institute reported that 34 per cent of university students “believe AI will eliminate jobs fast enough to cause civil unrest”. Given that AI is landing in a political climate turned febrile and populist, Kanagasooriam suggests, we may well see a party that “is very candidly Luddite”.
There is an opening here for a political entrepreneur. Polling for the Comment is Freed Substack suggests that 56-57 per cent those on the left and on the right see AI as a threat, as opposed to 41 per cent of centrists. So perhaps a single, self-styled anti-AI tribune of the people could draw together left-wingers angry at the role of autonomous weapons in Gaza and Palantir in the NHS with rural conservatives opposed to data centres, and angry unemployed graduates and their parents.
Like most I spoke to, Weiss thought the emergence of such a political force was likely – and added that it should happen, “because that is a growing and important voice and it needs to have adequate representation in the political sphere”. However, he doubted whether it would beat a counter-pitch promising to “reduce the cost of living by throwing the kitchen sink at AI”. Speaking while Starmer was still in power, he warned that Labour should avoid the temptations of stoking a “populist backlash”. If such a backlash comes, it is likely to remain a minority force, however consequential. But there is another factor that might unite a larger swathe of the public in opposition: an AI-induced calamity. As IPPR’s report puts it, “For now, anti-AI coalitions remain fragmented. But the state of AI in 2026 means this could change quickly, for example in response to a significant AI safety incident or a visible labour market shock.”
Joseph Miller of PauseAI recently visited Washington DC, to lobby politicians and AI companies. He found the general view was: “We obviously can’t do anything at this point. We’re just gonna have to wait for something to go horribly wrong.”
[Further reading: Will Andy Burnham call an early election?]
Content from our partners
Related
This article appears in the 20 Aug 2026 issue of the New Statesman, Summer Special
Join the debate
Subscribe here to commentThey knew what they were doing, the bros, when they rallied behind Trump. They could see the backlash coming – or, if you must, the ‘techlash’. No more Mr Nice Tech.
The funding of AI in the West is typical of Western Capitalism. During the introduction of the railways in England during the 19th century the government approved two companies to build a railway from London to the South Coast. The two bridges over the Thames were being built at the same time despite there being insufficient traffic to make both profitable. The private train companies were bankrupt by 1921.More than one of the current American AI companies will have to be taken over by its better organised competitors and quietly amalgamated into a competitive option. In the meantime the development of data centres and electricity supply options are being treated as if water and electricity were freely available in the quantities required. They are not available in any quantity imagined. Solar and Wind Turbine schemes are forced to queue for space in a National Grid system that has been ignored by politicians for a generation. Even Blair failed to penetrate the maze that the controllers of the National Grid maintained .