The machines are coming for your job. Not with clanking metal arms, but with silent lines of code. Large language models (LLMs) like ChatGPT are already reshaping white-collar work: drafting contracts, generating code, answering customer inquiries. In London, Canary Wharf's gleaming towers are quietly shedding junior analyst roles. In Manchester, legal firms are cutting paralegal headcount. The efficiency gains are undeniable. But as the government heralds a new era of productivity, a troubling question gnaws: who pays for the human cost? No one is raising the spectre of an automation tax. Yet the logic is brutally simple. If a company replaces a £40,000-a-year employee with a subscription costing £2,000, the state loses income tax, National Insurance contributions, and ultimately, that worker’s spending power. The firm pockets the profit; the public shoulders the burden.
Consider the numbers. HMRC data shows that income tax and National Insurance from employees accounted for £345 billion in 2023. If LLMs displace just 5% of the workforce over the next five years — a conservative estimate from the Office for Budget Responsibility — that’s a £17 billion hole in public finances. Meanwhile, corporation tax receipts are falling, not rising. The tech giants selling these tools pay minimal UK tax. Microsoft, which has invested billions in OpenAI, paid just £40 million in UK corporation tax last year on revenues of £12 billion.
The structural failure is obvious: the rewards of automation are privatised, the costs socialised. The Treasury’s own economic models assume frictionless re-employment, but that fantasy ignores the scarring effects of displacement. Older workers, those in administrative roles, and regions outside the South East will be hit hardest. Redundancy pay is a pittance; retraining schemes are patchy. The government’s AI Safety Summit was all about existential risk from superintelligence. Nobody mentioned the existential risk to a 52-year-old legal secretary in Leeds.
An automation tax would force firms to internalise these externalities. The concept isn’t new. Nobel laureate James Meade proposed a tax on robots in the 1970s. Bill Gates revived it in 2017. The mechanism could be simple: a levy on the cost savings from replacing workers with LLMs, or a direct payroll tax cut exemption for roles that cannot be automated. The revenue could fund universal basic skills credits or wage insurance for displaced workers.
Critics will scream innovation kill. They will say Britain cannot afford to hamstring its tech sector. But look at South Korea: they introduced a robot tax in 2018 by reducing investment tax credits for automation. Investment didn’t collapse. Or consider Denmark’s “flexicurity” model, which combines flexible hiring with generous unemployment benefits funded by employer contributions. Their economy thrives.
The alternative is worse. Unchecked automation will accelerate the hollowing out of the middle class. The Bank of England warns that up to 15 million jobs could be at risk. That isn’t a prediction; it’s a political choice. Without intervention, we hurtle towards a two-tier labour market: a tiny elite owning the AI, and a gig-economy mass competing for scraps.
The Treasury remains silent. The Department for Business and Trade is too busy wooing Silicon Valley. The Prime Minister is entranced by the promise of an “AI-powered Britain”. But when the tax revenues dry up and the dole queues lengthen, don’t say we weren’t warned. “It’s like watching a slow-motion train wreck,” says Dr. Emily Cox, an economist at the Institute for Fiscal Studies. “They’re not even looking at the tracks.”
This isn’t Luddism. It’s common sense. A tax on automation isn’t a tax on progress. It’s a price on the future. The question is whether Britain has the courage to charge it.
