This man just predicted when the UK will go to war – it’s closer than you think

TL;DR

Dr Keith Dear's AI firm Cassi forecasts a 25% chance of Britain in a major war before 2036.
Spending 3% of GDP on defence over five years could halve that risk to 10%, per his Commons testimony.
The approach beats big AI rivals by quantifying threats like Russia conflict (17%) to aid decisions.

The story at a glance

Chris Blackhurst profiles Dr Keith Dear, ex-RAF intelligence officer and Cassi co-founder, whose AI predicts UK war risks. Dear testified to the Commons Defence Committee, updating odds after US-Israel strikes on Iran. This comes amid worries over Britain's defence readiness.[[1]](https://www.independent.co.uk/voices/uk-world-war-ai-future-markets-britain-b2960068.html?loginSuccessful=true)[[2]](https://www.independent.co.uk/voices/uk-world-war-ai-future-markets-britain-b2960068.html)

Key points

Major war means over 500 British deaths or 250 in a conflict with 10,000 combatant deaths; last was Korean War.
Probability rose from 20% ("one in five") to 25% ("one in four") post-strikes on Iran.
17% chance of UK-Russia conflict before 2036, making it the likeliest major foe.
26% odds of China-Taiwan war, but only 11% boost to UK major involvement.
Cassi, an 11-person firm, tops AI forecasting rankings over Anthropic, Google, xAI, OpenAI, Meta.
Dear's background: 18 years RAF intelligence in Afghanistan, Iraq, Mali; ex-Downing Street adviser; Fujitsu role.
Firm aids governments, firms on geopolitics, markets, like 25% chance of Iran uprising by 30 April 2026.[[1]](https://www.independent.co.uk/voices/uk-world-war-ai-future-markets-britain-b2960068.html?loginSuccessful=true)

Details and context

Cassi uses an AI "Moneyball" method, inspired by Michael Lewis's book, to score and mimic top expert judgments where war data is scarce and misleading. Unlike sports, geopolitics lacks stable rules, so they test intuitions explicitly.

The firm helps Whitehall prioritise spending by quantifying collapse risks, like Treasury bids for extra funds. On markets, it spots non-consensus drivers like war or regulation over routine commodities.

Dear stresses numbers grab attention: vague terms like "substantial to severe" are ignored, but "60% to 80%" prompts action. Outcomes depend on users, akin to buying insurance despite 10% fire odds.

Key quotes

“The forecasts suggest that if the UK consistently spends 3 per cent of GDP on a five-year average basis by 2036, then the probability of the UK being involved in a major conflict drops by 50 per cent in relative terms: from 20 per cent to 10 per cent. A drop from one in five to one in 10.”[[1]](https://www.independent.co.uk/voices/uk-world-war-ai-future-markets-britain-b2960068.html?loginSuccessful=true)

“If you say the probability of something bad occurring has moved from ‘substantial’ to ‘severe’, nobody listens. If you say it’s gone from 60 per cent to 80 per cent or higher, then they do.” – Dr Keith Dear[[1]](https://www.independent.co.uk/voices/uk-world-war-ai-future-markets-britain-b2960068.html?loginSuccessful=true)

Why it matters

Rising war forecasts highlight UK's vulnerability amid global tensions like Iran and Russia. Leaders and investors gain tools to weigh defence spending against 50% risk cuts, while firms use probabilistic edges in uncertain markets. Watch if government hikes to 3% GDP or Cassi influences policy, though real-world tests remain uncertain.

LANG: en