IMF warns AI could wreck global financial system as 'arms race' accelerates At the IMF summit in Bangkok, officials are sounding the alarm about the potential for artificial intelligence to destabilize the global financial system. The rapid development of powerful AI tools has created an arms race, with cybercriminals and rogue states posing a significant threat to banking infrastructure, stock exchanges, and personal assets. Here in Bangkok it's monsoon season, and the tropical heat and humidity are making for uncomfortable conditions. But within the safely air-conditioned Queen Sirikit National Convention Centre, at the annual summit of the International Monetary Fund (IMF), something else is making attendees hot under the collar. Officials and regulators have been telling me one after another how terrified they are of the all-powerful AI tools being built at breakneck speed - and which could presage the meltdown of the global financial system. Britain may have once played a key role in the development of AI - through companies such as DeepMind, now owned by Google. But today the major AI players are all American or Chinese - while we have been pushed to the sidelines.London Financial Sector VulnerabilityMeanwhile, our economy is more vulnerable to this technology spiralling out of control than that of almost any other advanced nation. Together, the City of London and our financial and professional services sectors generate up to 12 per cent of national wealth. Any attack on them would be a direct attack on our resilience.Imagine what hackers or cyber-criminals, armed with AI tools such as Anthropic's unnervingly powerful Claude Mythos, could inflict on markets and systems in Canary Wharf As the world's second most important financial centre, challenged only by Wall Street, London hosts the biggest foreign exchange markets on the planet. It is home to a vast $4trillion market in 'complex interest rate derivatives', and is the centre of a payments system for some of the world's biggest banks - and much more.If even one of these networks were to be shut down by a rogue AI model, or one instructed by bad actors, it could cause havoc. This is not an idle fear. In the past 12 months alone we have seen terrible damage wrought on large British brands from Jaguar Land Rover to Marks & Spencer and most recently online fashion giant Asos.More frightening still is this week's bizarre hacking of several French supermarket chains, which saw refrigerators heat up to as much as 50C (122F) causing millions of euros worth of damage. Imagine then what hackers or cyber-criminals, armed with AI tools such as Anthropic's unnervingly powerful Claude Mythos, could inflict on markets and systems in Canary Wharf, the Square Mile or on your and my personal bank accounts, pensions and other assets.Payments System Attack RisksIf the payments system were attacked, ATMs could stop working, money transfers including mortgage payments be diverted, credit cards frozen, salary payments cancelled and much else. If the clearing systems used by all our High Street banks and building societies were disabled by a highly sophisticated - let alone superintelligent - AI, then millions of current and savings accounts could be drained in nanoseconds.Were the London Stock Exchange to be targeted, then personal savings or pension funds' investments in stocks could be wiped out, causing the market to crash. An arms race is accelerating. Cyber-security firms and others are trying to safeguard the machinery of the global financial system - while, on the other side, rogue states such as North Korea and Iran are hell-bent on causing chaos and division and stealing anything they can get their hands on.As the IMF warns, the world's ability to combat such threats is becoming ever more difficult because of the near-instantaneous nature of attacks enabled by AI. In more than 70 per cent of reported cyber-attacks so far in 2026, the providers of the software designed to repel them were unable to update their systems before the inadequacies were exploited by hostile forces. Back in April, I spoke to the Governor of the Bank of England, Andrew Bailey, in Washington DC.Bank of England Governor WarningHe alerted me even then to the dangers that powerful new AI tools represented to the financial system. His anxiety was palpable and prescient. Six months on, many of the finest minds in global finance have now joined him in focusing on AI - both its potential for money-making and its inherent risks. Yes, AI can be a force for economic good: I have followed closely how it has bolstered the economies of chip and hardware manufacturers in Asia.But the speed of its development - which could soon become 'recursive', meaning the AI models improve themselves by continually refining their own computer code - is now regarded as the primary risk facing the international economy. That is all worrying enough. But at the same time, this potentially disastrous technology is being built on sand: specifically the mountain of debt on which the industry's mad rush for data centres and other commercial advantage is being built.