British Foreign Secretary Yvette Cooper has warned against allowing artificial intelligence to develop further without international rules. She used the dramatic image of Hiroshima, where the world’s first atomic bomb was used as a weapon of war. In the case of nuclear weapons, Cooper argued, international controls began only after the world had seen the destructive power of the new technology. With AI, she warned, the world must not wait for a similar shock.
She set out the argument in an essay for Chatham House and in an interview.
Cooper wants global rules for AI. These should be developed with the explicit involvement of the US and China. For her, AI is no longer merely a technological or economic issue, but a central question of foreign and security policy. She predicts that AI will become one of the dominant issues in British foreign policy over the next two years. She also warns of hostile use by states, state-linked criminal groups, extremists and terrorists.
The Risks Are Already Here
These are not science-fiction scenarios. Cooper is pointing to concrete risks: cyberattacks, hybrid operations, disinformation, automated surveillance, military applications and the acceleration of criminal capabilities.
She also places AI within a wider global security context. Alongside AI, she names the contested issue of climate change, irregular migration and foreign interference as threats to liberal democracies.
In Cooper’s view, Britain and Europe must prepare more seriously for a world in which the US no longer automatically plays the role of global arbiter and security guarantor. She is calling for a closer and more stable British-European security architecture.
Rules Before the Shock
The rules, Cooper argues, must come before a catastrophe, not after one. The Hiroshima comparison is deliberately stark. AI is far from being an atomic bomb. But its risk lies in its ability to create strategic realities faster than politics, law and international institutions can respond.
Cooper’s warning that AI should be placed in the same strategic category as nuclear weapons, cyberwarfare or biotechnology reflects the fact that it is a key technology capable of changing power relations, warfare, domestic security and freedom. The foreign secretary’s demand is that international guardrails be put in place before a major incident dictates the terms.
A Moratorium
Cooper is not alone in calling for an AI moratorium or tighter rules. Some AI companies have also urged stronger safeguards. UN Secretary General António Guterres recently warned at a conference in Geneva that a globally coordinated framework was needed to minimize potential risks. AI had the potential to transform economies, reshape labor markets and influence elections and global security. Innovation needed guardrails, Guterres argued. If AI was to be powerful, it had to be governed.
US Senator Bernie Sanders and Representative Alexandria Ocasio-Cortez also introduced the Artificial Intelligence Data Center Moratorium Act in March 2026. In an open letter published by the Future of Life Institute in March 2023, the signatories called for a pause of at least six months in the training of AI systems more powerful than GPT-4.
The signatories included Elon Musk, Steve Wozniak, Yoshua Bengio, Stuart Russell and many other researchers and entrepreneurs. Their motivation was certainly different from Cooper’s military focus, but it too rested on the assumption that the dangers were serious.
The London-based think tank Chatham House argued in March 2026 that international AI governance was at an impasse. Geopolitical tensions, weak institutions and the power imbalance between states and private corporations were making binding rules more difficult. The authors warned that real progress might become politically possible only after a crisis. That is precisely the pattern her appeal is meant to break.
AI Beyond Control
Another Chatham House paper, published in April 2026, described how deeply AI is already becoming intertwined with defense, security and sovereignty policy. The authors saw a shift toward a more fragmented AI landscape, increasingly shaped by security concerns. States and private technology companies were moving closer together.
The threat posed by AI is twofold: misuse by humans for criminal purposes and the possibility of artificial intelligence operating beyond human control. The more AI operates in autonomous weapons systems and is able to carry out offensive and defensive measures without human intervention, the greater the risk of such loss of control.
There are also hybrid threats, since AI can be used directly against the infrastructure of foreign countries. The similarity with a nuclear weapon lies in the fact that, once unleashed, AI may in some cases be no easier to stop than a chain reaction already under way in a nuclear device. Some AI systems have already shown an ability to resist commands to delete themselves.
There has so far been no international response to Cooper’s intervention. The unanswered question is how to prevent governments from using regulation to cut large parts of the population off from legitimate and even experimental uses of AI.