UK Foreign Secretary Warns of 'AI Hiroshima' Without Global Safeguards
In brief
- Cooper urges preemptive AI governance before crisis forces international action
- Early nuclear agreements emerged only after Hiroshima; AI safety pacts must come first
- UK AI Security Institute flags rapid advances in AI cyberattack capabilities
- Anthropic CEO demands mandatory third-party testing of frontier AI models
The Case for Preemptive Action
Cooper said AI offers breakthroughs but also presents new risks as the technology becomes more powerful and widely available. The same technologies reshaping healthcare are also reshaping the future of warfare, crime, and social cohesion in alarming ways. Her warning reflects a growing consensus among policymakers that the window for proactive governance is narrowing.
Cooper compared the current race to develop AI systems with the early nuclear arms race. Global safety agreements on nuclear weapons emerged only after the world witnessed the devastation at Hiroshima. She's arguing that waiting for an AI-driven crisis before establishing shared safety principles would repeat that costly mistake.
Concrete Threats on the Horizon
Recent developments underscore the urgency. In May, the UK's AI Security Institute warned that rapid gains in AI cybersecurity abilities occurred after OpenAI's GPT-5.5 became the second model to complete a simulated cyberattack without human assistance. The International Monetary Fund warned that AI could amplify cyberattacks against the global financial system by lowering the skill needed to exploit vulnerabilities.
These aren't theoretical risks. They're emerging capabilities that demand coordination among major powers.
Path Forward
Cooper called on Britain to use its diplomatic influence to bring together the United States, China, and other major AI powers to establish shared safety principles and standards. She pointed to the 2023 AI Safety Summit at Bletchley Park, where global leaders from 29 countries and the European Union met to discuss the emerging risk of AI, as evidence that multilateral engagement is possible.
Not all policymakers agree on speed. In June, President Donald Trump signed an executive order creating a voluntary framework for reviewing advanced AI models before release. But Anthropic CEO Dario Amodei argued that transparency requirements are no longer enough and called for mandatory third-party testing of frontier models. The debate over how hard to push is intensifying.
