Planning for the Worst Case: UK's AI Minister Says a Jobs Contingency Plan Is Coming

At a fringe meeting at the Labour Party conference in Liverpool, Britain’s Minister for Artificial Intelligence said something governments rarely say out loud: it is preparing for the possibility that AI goes badly wrong for the labour market — and it wants to be ready before it happens.

Kanishka Narayan, appointed AI minister in July 2026, told the meeting that one of his top priorities is “thinking through what you would do in contingency” if AI-related job losses turn out to be as severe as the more pessimistic forecasts predict. The government must “take seriously the possibility of an unprecedented impact on the jobs market,” he said, adding: “If it’s going to hit us in a big way, you are going to want to be as ahead of the curve as possible.”

It is a striking posture for a government that has spent two years pitching itself as the most pro-AI administration in Europe. But it is also a sign of how fast the political weather has changed. The same conference heard Foreign Secretary Ed Miliband fire a warning shot at “tech titans,” arguing that “the lesson of history is that we cannot leave it to corporations to put in place the guardrails to serve the public interest.”

The numbers behind the warning

The starkest figures at the fringe meeting came from Carsten Jung of the IPPR think tank, who is leading work on what he described as a “pandemic preparedness”-style report on AI labour market disruption.

• 8 million jobs — Jung’s worst-case estimate for the number of UK jobs that could be “negatively affected” by AI.
• 11% of UK jobs are currently “highly exposed” to replacement by AI chatbots, “led by secretarial and administrative occupations.”
• 60% of tasks across the UK economy could become exposed as more sophisticated agentic AI — systems that act autonomously rather than just answering questions — spreads through the economy.
• 40% higher exposure for women than men, because a higher proportion of women work in service-sector roles most vulnerable to automation.

Jung was candid that IPPR’s own “stark scenarios” published in 2024 were “to some extent” wrong — but in an uncomfortable direction. AI has since evolved from automating “boring” routine tasks to handling “more high-level reasoning and creative tasks,” a shift that widens the blast radius far beyond the clerical roles the original analysis targeted. “We are not preparing enough for something quite big that’s coming,” he warned.

Why the official data looks too calm

One of the most interesting parts of Narayan’s contribution was his explanation for why current labour statistics show so little AI impact. Firms report “no aggregate impact in the overall labour market today” — but the minister suspects the truth is hiding in plain sight, through what he called the “silent adoption” of AI by workers that never shows up in industry surveys.

His evidence: one tech firm told him directly that it had hired fewer people because tasks had been automated. Hiring that never happens is invisible in unemployment data. The “actual net impact might not be fully captured,” Narayan acknowledged — a refreshingly precise admission from a serving minister, and one that squares with Morgan Stanley research from earlier this year suggesting AI-related job cuts have been landing hardest in Britain among major economies.

The implication for policy is significant. If displacement happens primarily through reduced hiring rather than layoffs, the pain arrives slowly, unevenly, and mostly among new entrants to the labour market — young people who never get the first rung of the career ladder.

What a contingency plan might contain

Neither Narayan nor Jung sketched a finished blueprint, but the fringe meeting pointed at several directions.

Retraining at scale. Jung argued the government must be ready to help retrain service-sector workers, who are disproportionately exposed in the UK’s service-heavy economy.

Tax and transition management. He suggested redesigning tax structures to “make it a bit more attractive to hold on to workers than automating them” — a carbon-pricing-style nudge applied to labour substitution — and to “slow down the transition slightly” so adjustment happens at a pace society can absorb.

Fiscal honesty. Jung was blunt that “a lot of the policies we might want to do in a really severe disruption scenario will cost a lot of money,” particularly for the UK, which is more service-based than other G7 nations. A genuine contingency fund is a budget question, not a press release.

Worker voice and transparency. Narayan promised “much better transparency when AI and technology is adopted” so productivity gains can actually be measured, and said he is determined that workers have a voice in the transition. Mike Clancy, general secretary of the Prospect trade union, welcomed the commitment but warned against “creating the conditions in which working people become victims” of AI.

Regulation as a dial, not a switch. The minister hinted the contingency planning could extend to “labour market policy” and “even where we might go on some aspects of regulation” — the first explicit signal that the UK’s light-touch AI framework could be tightened if employment data turns ugly.

The bigger picture: from sprint to safety

Narayan’s contingency talk did not come out of nowhere. It lands in a week when the UK’s AI policy posture has visibly shifted toward the safety end of the spectrum. Miliband’s speech pledged to use the UK’s G20 presidency next year to “lead the efforts to manage AI so we keep our country and our world safe,” and explicitly rejected the idea that tech companies can self-regulate: “It is neither their purpose nor their motive, theirs is to maximise profits.”

For a minister who has spent his first months in office championing AI adoption — “AI kills jobs, creates jobs and you just have to manage the transition,” he told the Financial Times in June — the new emphasis on worst-case planning marks a deliberate rebalancing. The government is not abandoning its pro-innovation stance. It is buying insurance.

Why this matters beyond Britain

The UK is running an experiment the rest of the world is watching. It is a large, service-dominated G7 economy with an unusually centralised AI policy apparatus — a dedicated AI minister attending cabinet, a safety institute, and now an explicit contingency-planning mandate. If AI-driven displacement does accelerate, Britain will feel it earlier and more sharply than economies with larger manufacturing bases, and its response will be the template other governments copy.

The IPPR’s full “pandemic preparedness” report on AI labour disruption is still in the works. When it lands, expect the contingency debate to move from fringe meetings to the front bench — and possibly into the next Budget. The quiet admission from Liverpool is that the age of pretending AI is purely a jobs-creation story is over, at least in one corner of Westminster.

Sources for this article include the BBC’s report from the Labour Party conference in Liverpool and IPPR’s published analysis of AI labour market exposure.