The G20 has identified cyber threats posed by AI as a key danger to the global financial system

Neural networks pose an increasing threat to the global financial system, and security measures surrounding their release must be a priority, wrote Andrew Bailey, chairman of the Financial Stability Board and governor of the Bank of England, to the heads of central banks and financial authorities of the G20 countries ahead of their meeting in the U.S. on August 31, according to The Wall Street Journal.

"The emergence of the latest AI models has complicated the situation: they are becoming increasingly autonomous, better at solving complex problems, and their capabilities for carrying out attacks are growing," the WSJ quotes Bailey as saying in a letter.

A disruption in one country could affect financial institutions in others: many of them rely on the same technology providers and shared infrastructure, Bailey noted. Differences in legislation, levels of cybersecurity, system resilience, and the ability to recover from attacks pose an additional risk, he warned.

Bailey called for special attention to be paid to ensuring safety when releasing new AI models. “Many countries lack procedures governing the development, release, and use of advanced AI models. This poses risks not only to the financial sector,” the WSJ quoted him as saying.

Banks and other financial institutions should prepare for “more severe scenarios” in which disruptions affect several companies at once or the technology platform they all share, the economist warned.

Regulators fear that AI could uncover unknown vulnerabilities in financial institutions’ systems and quickly circumvent new security measures. The European Central Bank has required eurozone banks to submit a plan by October 31 to protect against threats posed by these new models. In an open letter, leading Western AI developers called on authorities and the business community to urgently address the protection of critical infrastructure against AI-powered cyberattacks, which “will become much more widespread and sophisticated in the coming months.”

But the potential consequences for the markets are not limited to cyberattacks. Bailey also warned that if the boom in AI investments, fueled by debt, gives way to a downturn, stock markets could crash, according to The Telegraph.

These risks are exacerbated by the fact that investors are borrowing huge sums and investing them in a small number of AI companies and data center operators, driving their valuations to “astronomical” levels. According to Bailey, this overvaluation of AI assets could be one of the factors contributing to a sharp economic downturn.

“Markets remain at risk of a sharp and uncontrolled correction that could affect various countries,” Bailey said. The problem, he said, is not only that investors have started borrowing more: leveraged investments are concentrated in a small number of AI companies and data center operators, whose valuations have skyrocketed to astronomical levels.

These risks are exacerbated by the vulnerability of government debt markets, notes The Telegraph. The cost of borrowing worldwide has risen recently, most notably in the U.S., where the national debt exceeded $40 trillion in August. According to Bailey, the growing role of hedge funds—which use leverage while simultaneously holding government bonds—increases the risk that problems in the stock market will affect the government debt market—or vice versa.

This article was AI-translated and verified by a human editor