Starmer risks a major war within 10 years without one change, warns AI | UK | News
An artificial intelligence system billing itself as the most accurate forecaster on earth has put Britain’s chances of fighting a major war within the next decade at one in four — and says only a dramatic increase in defence spending can bring that risk down.
Cassi AI, whose founder claims it produces forecasts “just off the best forecast that humanity can make under uncertainty”, put the odds of UK involvement in a major conflict at one in five before the Iran war broke out on February 28. That figure has since climbed to 24 per cent.
The system’s modelling suggests a single policy decision could transform the picture entirely. Were Britain to commit to spending 3 per cent of GDP on defence averaged over five years and hit that target by 2036, the conflict probability would fall to just 5 per cent — one in twenty.
What qualifies as a major conflict
For Cassi’s purposes, a major conflict is one in which British fatalities hit 500, or 250 in a war that claims 10,000 combatant lives in total. No conflict involving Britain has crossed that threshold since the Korean War.
The insurance question
Dr Keith Dear, an RAF intelligence officer for 18 years who went on to found Cassi, put the stakes in stark terms, according to a Times report: “If you told me there was a one-in-four chance that my house would burn down, I would take out insurance.
“The question this poses to politicians is: is it worth spending 3 per cent on defence to save those lives and the obvious crippling costs of a war?”
Dear, a former No 10 foreign policy adviser, said the system outperforms every major AI rival on accuracy — sitting above xAI, OpenAI, Anthropic and Google on a forecasting leaderboard.
Rather than relying on a single model, Cassi operates like an intelligence cell: gathering data from multiple sources, running it through separate analytical strands and weighing up the outputs of competing AI systems before reaching a conclusion.
Where spending stands
Britain is currently committing roughly 2.4 per cent of GDP to defence. Starmer has pledged to push that to 2.5 per cent by April next year, with 3 per cent pencilled in for some point in the following parliament.
Cassi’s modelling gives only a one-in-four chance of the 3 per cent target being reached by 2036. Miss that window and the risk reduction is significantly smaller, the forecast concludes.
The AI also examined how the trajectory of the Ukraine war could shape Britain’s exposure to conflict. An early end — before 2027 — would nudge the UK’s conflict probability upward by 5 per cent. A prolonged war running into 2029 would push it up by 15 per cent.
There is, however, a narrow path through. Should the Ukraine war wind down somewhere between January 1, 2027 and December 31, 2028, Cassi’s modelling suggests Britain’s conflict risk may remain broadly unchanged.
