Global cooperation needed to tackle AI threats, says Bank of England governor
Source Entity
Kalyeena Makortoff and Heather Stewart
Bank of England Governor Andrew Bailey has urged for international cooperation to manage AI-related threats, arguing that the US and the Trump administration cannot address these challenges in isolation, particularly following the restriction of Anthropic's Claude Mythos model for foreign users.
The Imperative for Global AI Governance
In a significant diplomatic and economic intervention, Bank of England Governor Andrew Bailey has sounded an alarm regarding the escalating risks associated with artificial intelligence. Bailey's core thesis is that the current trajectory of AI development, characterized by nationalistic competition and restrictive access, is insufficient to mitigate systemic global threats. By explicitly stating that the United States and the Trump administration cannot achieve their safety and security ambitions alone, Bailey is highlighting a critical vulnerability in the current geopolitical landscape: AI does not respect national borders, and therefore, a unilateral approach to its regulation is fundamentally flawed.
The Friction of AI Protectionism
Central to this discussion is the recent decision by the US government to temporarily ban foreign nationals from accessing Anthropic’s 'Claude Mythos' model. This move represents a shift toward 'AI protectionism,' where frontier models are treated as strategic national assets rather than global utilities. While the Trump administration may view such restrictions as a necessary measure to prevent adversarial states from leveraging high-end AI for cyber warfare or economic espionage, Bailey suggests that this isolationist strategy may be counterproductive. When the world's most powerful models are siloed, the global community loses the ability to create shared safety standards and collaborative monitoring systems, potentially pushing other nations to develop their own, less-regulated alternatives in a desperate race for parity.
Financial Stability and Systemic Risk
As the head of one of the world's most influential central banks, Andrew Bailey's concerns are deeply rooted in financial stability. The integration of AI into global trading, risk assessment, and monetary policy introduces a layer of complexity that can lead to 'flash crashes' or systemic failures if not governed by a coherent set of international rules. If AI threats—such as automated market manipulation or AI-driven financial fraud—are handled on a country-by-country basis, the resulting regulatory arbitrage could allow risks to migrate to the least regulated jurisdictions, eventually spilling over into the global economy. Bailey is essentially arguing that AI safety is not just a technical or political issue, but a macro-prudential one.
The 'Frontier Model' Dilemma
The mention of the Claude Mythos model underscores the growing divide between 'frontier AI' (the most advanced, general-purpose models) and specialized AI. The ability of these models to reason across multiple domains makes them dual-use technologies—capable of both immense productivity gains and immense destruction. By restricting access, the US may be attempting to maintain a 'technological moat.' However, the history of nuclear proliferation suggests that secrecy and restriction often accelerate clandestine development elsewhere. The 'Mythos' incident serves as a catalyst for Bailey's argument that only a transparent, multilateral framework can ensure that these models are developed with safety guardrails that are globally recognized and enforced.
Predicting the Shift Toward Multilateralism
Looking forward, it is likely that the tension between US-led restrictions and the call for global cooperation will lead to the formation of an international AI regulatory body, similar to the International Atomic Energy Agency (IAEA). As the risks associated with AI become more tangible—ranging from autonomous weapon systems to the destabilization of labor markets—the pressure on the US to pivot from a policy of restriction to one of leadership-through-collaboration will increase. We can expect to see the emergence of 'AI Treaties' where nations agree on minimum safety benchmarks in exchange for shared access to compute power and cutting-edge models.
Conclusion: Diplomacy Over Isolation
Ultimately, Andrew Bailey's warnings serve as a reminder that in the age of intelligence, isolationism is a luxury that global powers cannot afford. The 'Claude Mythos' ban may provide a short-term tactical advantage for the United States, but it does nothing to solve the existential risks that AI poses to humanity as a whole. For the global economy to remain stable and for AI to be a force for good, the transition from nationalistic competition to strategic cooperation is not merely an ideal—it is a necessity for survival.