We are ignoring all the warnings about AI – and heading towards disaster
June 2025, King's Cross: Campaigners from 'Pause AI' and AI professionals stage a protest outside Google DeepMind offices | Image by: Vuk Valcic / Alamy
4 min read
We are racing headlong into an AI-driven future without basic safeguards in place
In the 1980s, straight out of film school, I directed a current affairs programme about the City of London’s ‘big bang’, during which a senior US Bank executive told us that deregulation would “end in disaster”, before adding, almost cheerfully, “but meanwhile, there’s so much money to be made”.
Deregulation supercharged speculation and excess, during which – as a beneficiary once admitted – “we just had to stand downwind to make eyewatering pots of money”. That story ended in the 2008 financial crash. Millions lost their homes and jobs, public services were hollowed out, and taxpayers were left to clean up the mess. The gains remained private; the losses were public. The warnings had been given. They were ignored.
They are being ignored again. We are now racing headlong into an AI-driven future without basic safeguards in place. A handful of companies, largely in Silicon Valley and China, are building systems on data harvested from billions of people and monetising them on a global scale with no regard for the damage that could follow. The same mantra – that regulation is a barrier to innovation – is once again exhorted by governments and the technology sector.
In opaque deals, AI companies are being given access to broadly defined ‘government data’ without democratic scrutiny – even though, once surrendered, they cannot be retrieved. Among the jewels of the UK’s unique data sets are those from schoolchildren, CCTV footage and our cradle-to-grave NHS data, the most valuable health data set in the world. These are sovereign assets that could shape the economic and political future of the next generation. But decisions about who can access UK data sets and what value they have are being taken behind closed doors.
AI companies are being given access to ‘government data’ without democratic scrutiny
It is not only personal and public data that is in play. In their eagerness to please Silicon Valley and Washington, ministers have been prepared to give away creative property that does not belong to them – against the will of its owners and in the name of ‘AI innovation’, the government has floated exceptions to copyright law that would allow tech companies to scrape and exploit creative work without permission or payment. In any other context, this would be called stealing.
Meanwhile, the risks are mounting. In bedrooms across the country, children are increasingly turning to AI chatbots that give illegal advice and create emotional dependency. In the City, a senior partner at a major investment firm recently said “at least one” of the leading AI companies will fail. “There isn’t enough money in the world to get a return on all that investment.” In a gobsmacking irony, last week the UK’s National Cyber Security Centre, in reporting on serious cyber-attacks, said: “From local coffee shops to providers of critical national infrastructure, every organisation must understand their exposure, build their defences, and have a plan for how they would continue to operate without their IT”. The efficiencies promised by AI come with caveats and vulnerabilities of their own.
AI’s consequences – from biased algorithms and copyright theft to systemic shocks and industrial disruption – are foreseeable, predictable and, if we act, preventable. Only last month, Turing Award winners Geoffrey Hinton and Yoshua Bengio were among 200 signatories of a letter begging for a pause and for guardrails.
We know how the story ends: the financial crash; Boeing 737 Max; the explosion of online harm. In every case, warnings were dismissed, profits prioritised, and the public bore the cost. If we want a different ending, we need to heed the voices calling for democratically accountable technology – innovators, parliamentarians, experts and citizens – who are not anti-tech but pro-accountability.
Baroness Kidron is a Crossbench peer
