Whitehall is about to get a new breed of bureaucrat. Every government department will be required to appoint a Chief Artificial Intelligence Officer, a senior role tasked with embedding machine learning across policymaking and public services. The mandate, confirmed by the Cabinet Office this morning, is the most aggressive push yet to transform British governance through artificial intelligence.

The news landed at 7am. By 8am, the civil service unions were already raising concerns about a shortage of qualified candidates. “We have maybe a dozen people in the entire country with the right combination of technical expertise and Whitehall experience,” a senior official told me, speaking on condition of anonymity. “Now we need dozens overnight.”

The role is not optional. Each department must have a CAIO in place by April 2025. The appointee will report directly to the permanent secretary and sit on the departmental board. Their brief is broad: oversee the ethical use of AI, identify areas for automation, and ensure compliance with new regulatory frameworks.

But here is the tension. Whitehall has a poor track record with big technology projects. The £10 billion NHS IT system was scrapped. The Universal Credit rollout was plagued with delays. Now the government is betting that AI can save billions by streamlining everything from benefits processing to border control.

I spoke to a data scientist who recently left the Office for National Statistics. “The problem isn’t the technology. It’s the culture. Senior civil servants don’t understand data, and they don’t trust algorithms. They want the efficiency, but they’re terrified of the headlines when an AI denies someone a pension.”

That fear is not unfounded. Last year, the Home Office faced backlash after an algorithm used to flag visa applications was found to be biased against certain nationalities. The new CAIOs will be responsible for preventing such scandals, but they will also be racing against time. The Treasury is reportedly pushing for rapid deployment to cut costs before the next spending review.

The mandate comes with a central support unit. A new Office for AI Adoption will sit inside the Department for Science, Innovation and Technology, offering shared tools and training. It will be led by Priya Patel, a former DeepMind policy director. She told me the goal is to avoid a patchwork of different systems. “We need interoperability. The Department for Work and Pensions can’t use a model that the Home Office rejects.”

But each department will have its own budget. That creates a risk: the richest departments will hire the best talent, while others are left with second-tier appointments. A source in the Treasury acknowledged this. “We are looking at a central recruitment pool, but the details are still being worked out.”

Meanwhile, the opposition is watching closely. Labour’s shadow science secretary said the plan was “underfunded and overambitious”. He pointed to the lack of a dedicated AI bill in the King’s Speech. “They want to appoint officers without a legal framework for accountability. That is a recipe for disaster.”

The Cabinet Office insists the opposite is true. The CAIOs, they argue, will provide the human oversight that algorithms currently lack. “This is not about machines running the country,” a spokesperson said. “It’s about putting experts in place to ensure we use these tools safely and effectively.”

Yet the speed of the rollout raises questions. How do you find 200 experts in 18 months? And once you find them, how do you integrate them into a system that rewards caution and penalises failure? One former permanent secretary put it bluntly: “Whitehall eats clever people for breakfast. These CAIOs will either be the most powerful officials in the room, or they’ll be cut out of the loop within a month.”

The first appointments are expected within weeks. The Department for Health and Social Care has already advertised for its CAIO, offering a salary of £140,000. That is more than a senior civil servant, but less than a junior data scientist at Google. The market is tight.

One thing is certain: the language of Whitehall is about to change. Algorithms, neural networks, large language models. These will become the new jargon of the corridors of power. Whether they will also become the tools of better government remains to be seen.