UK Foreign Secretary Warns of an "AI Hiroshima" Without Rules

The UK's top diplomat has a blunt message for world governments: don't wait until something goes catastrophically wrong with AI before agreeing on how to control it.

In an essay for the Chatham House think tank, Foreign Secretary Yvette Cooper draws a direct comparison to the birth of the nuclear age — pointing out that meaningful international agreement on nuclear weapons only came together after the world witnessed exactly how destructive they were, at Hiroshima. Her argument is that AI shouldn't get the same free pass. Cooper frames AI risk as one of the defining security challenges of the next decade, and possibly the single biggest one — one she expects to dominate foreign policy conversations for years, not months.

She's not arguing AI is all downside. During a trip to Shenzhen, Cooper saw firsthand how AI and robotics are already being deployed in healthcare, and she acknowledges the upside is real. Her concern is narrower and sharper: the same leap in capability that's improving medicine is also reshaping what warfare, organized crime, and threats to social stability look like — and doing it faster than governments are used to responding.

Cooper also widens the lens beyond AI specifically, describing a world where danger no longer comes only from traditional battlefields. Hybrid threats — cyberattacks and disinformation campaigns launched from the other side of the planet — can disrupt a country's companies and infrastructure overnight, without a single soldier crossing a border. She ties cybersecurity directly to national, economic, and democratic stability, and says the UK needs to get better at defending against exactly this kind of hybrid pressure. She specifically references expanding Britain's ability to identify, publicly call out, and sanction Russian disinformation operations.

Her bottom line is essentially a plea for getting ahead of the problem instead of reacting to it after the fact. With nuclear weapons, the world only built shared safety norms after seeing the damage firsthand. With AI, Cooper argues, there's still a window to agree on principles, standards, and limits before that kind of wake-up call happens — and she's specifically calling for the US, China, and other AI superpowers to be part of that conversation, not just the UK and its usual allies.

Whether major AI powers actually sit down together on this, given how competitive the AI race already is, remains the open question her essay doesn't really answer.