Home / Publications / Blog / ForIntel Research / The UK Is Building a Contingency Plan for AI Job Losses Before They Happen

The UK Is Building a Contingency Plan for AI Job Losses Before They Happen

UK AI Minister Kanishka Narayan says the government is preparing for a severe wave of AI job losses, citing IPPR's worst-case estimate of 8 million UK jobs at risk and a run of 2026 layoffs at Reach, Oracle, Amazon, Meta, and Microsoft.

By Foragentis teamPublished 2026-10-022 min read

The UK's AI Minister, Kanishka Narayan, said this week that the government is drawing up contingency plans for what he called potential "unprecedented" job losses from AI. He described the approach as "pandemic preparedness" for the labour market: plan for the disruption before it lands, instead of reacting once it has already cost people their jobs.

The number behind the urgency comes from the Institute for Public Policy Research. IPPR's research puts the worst-case scenario at roughly 8 million UK jobs affected by AI automation, with no offsetting GDP gains to show for it. The same research estimates about 11% of UK worker tasks are already exposed to automation in this first wave of generative AI adoption, a share IPPR says could climb to 59% in a second wave. Administrative roles are the most exposed today.

The backdrop makes the number harder to wave off. UK unemployment stood at 4.9% in the three months to July 2026, up 0.2 percentage points on the year, with payrolled employees down by roughly 100,000 over the same twelve months. And the layoffs piling up right now have an unusually direct line to AI spending: Reach, the newspaper group behind the Mirror and the Express, cut 220 editorial jobs. Oracle cut US cloud infrastructure staff even as it poured billions into AI capacity. Amazon, Meta, and Microsoft each announced thousands of job losses in the same window. Companies are cutting headcount and increasing AI investment at the same time, which is exactly the pattern a "pandemic preparedness" framing is built to catch early.

Narayan's proposed response leans on visibility rather than a ban: "much better transparency when AI and technology is adopted," and more say for workers when new technology changes their jobs. That is a narrower ask than it sounds. It does not require predicting which jobs disappear. It requires knowing, in something close to real time, where AI is actually being deployed and what it is actually displacing, instead of finding out from a layoff announcement after the fact.

That is the part worth sitting with. A contingency plan is only as good as the evidence feeding it. A government, or a company, trying to prepare for labour market disruption needs the same thing at every scale: current, sourced information about where a technology shift is actually landing, not a lagging indicator from three months of unemployment data. That is the same instinct behind building market intelligence that tracks demand and workforce shifts as they happen rather than after the quarterly numbers come out. A closer look: https://foragentis.com

A chatbot hallucinated nuclear weapons cargo on a Chinese ship — troops nearly boarded it (CNN exclusive)

CNN's exclusive reporting: during the U.S.-Iran war, an AI chatbot hallucinated nuclear cargo on a Chinese vessel, and the fabricated report nearly triggered a military boarding. The failure traces to one missing engineering requirement.

An Appeals Court Just Issued the First Ruling on AI Training and Fair Use. The AI Company Lost.

A US appeals court upheld Thomson Reuters' copyright win over Ross Intelligence on September 29, 2026, the first appellate decision on AI training and fair use. Here is what it actually decided, and why it matters even if your AI isn't a legal-search tool.

Unsealed NYT v. Microsoft/OpenAI filings: Microsoft exec called AI training scraping the largest theft of labor in human history

Newly unsealed filings in The New York Times's copyright suit against Microsoft and OpenAI include a Microsoft scientist's own words: AI training-data scraping as "the largest theft of labor in human history." The admission points at a fixable design flaw.

AI rights movement (Ufair/Samadi) vs. Suleyman's Seemingly Conscious AI warning

The Guardian's profile of the AI rights movement puts consciousness on trial. The more useful question it raises: when should a fluent AI answer count as evidence at all?