Major Banks on High Alert for New AI-Driven Catastrophe
The banking industry is currently gripped by anxiety regarding the rapid advancement of Artificial Intelligence (AI) technology. The primary trigger is the emergence of Anthropic’s ‘Mythos’ model, which can easily detect system vulnerabilities.
Banking institutions worldwide fear that such sophisticated systems will be exploited by cybercriminals to launch faster and easier attacks. Consequently, banks holding vast amounts of sensitive data and significant funds could become prime targets for breaches.
Several nations have already urged the banking industry to strengthen their defence systems and enhance user privacy security. Recently, bankers from major institutions gathered again at the annual ECB conference in Portugal, with the primary focus being the impact of AI.
The consensus from these discussions concluded that AI possesses the capability to disrupt all sectors and create unprecedented problems. These include impacts on financial markets, the labour market, bank lending, security, and even electricity demand.
“If AI delivers results that exceed expectations, it will impact financial stability. If AI delivers results below expectations, it will also impact financial stability,” stated Torpend Slok of Apollo Global Management to global interest rate policymakers during a keynote panel session in the Sintra resort, as reported by Reuters.
AI became such a dominant theme in Sintra that the topic permeated every discussion, ranging from immigration and surveillance to climate change.
Indeed, AI overshadowed the new Federal Reserve Chair, Kevin Warsh, during his debut meeting with fellow central bankers, becoming the star attraction of the three-day event.
While AI can enhance every aspect of life, many speakers expressed concern that AI could also disrupt lives, sometimes illegally, and that financial officials possess little to no tools to stop it.
“I believe this is the most defining period for our respective economies in our lifetimes,” said Warsh regarding the AI revolution.
“Who would have thought that when the internet was born, that technology would create 1.5 million jobs as Uber drivers? We are only in the early stages of this revolution,” he remarked during the ECB forum.
Risks of Disaster Ahead
In the trading world, automation already performs most functions. However, the momentum driven by AI could create market bubbles with extraordinary speed, subsequently triggering their collapse. This could enable profit-making during both price rises and falls through forms of collusion that are now illegal.
“Something even more sophisticated and potentially more unsettling is the ability of these algorithms to coordinate in manipulating price movements,” said Itay Goldstein, a professor from the University of Pennsylvania.
“These algorithms are indeed capable of manipulation to create bubbles that lead to market crashes. I believe this has more significant implications for financial stability,” he added.
One potential bubble already created by AI is the AI stock bubble, partially driven by massive capital expenditure on AI building blocks, which Slok estimates has added one percentage point to US GDP alone.
Although valuations have declined in recent weeks, experts have compared the rapid price increases to some of the largest asset price declines in history, such as the British railway mania of the 1840s, the Roaring Twenties, or the dot-com bubble.
“The scale and speed of the current AI investment explosion, accompanied by high expectations for productivity rewards, resembles these precedents, highlighting the potential risk of a downturn in the short term,” stated the Bank for International Settlements in a report.
Facilitating yet Complicating
AI is also assessed to both assist and complicate the lending process. Banks will be able to perform more sophisticated credit analyses and provide funds to borrowers previously beyond their traditional reach. However, oversight of this will become a significant challenge.
“How will regulators assess lending decisions made autonomously by AI systems? The process resembles a ‘black box’ that is difficult to understand. There is a potential lack of transparency or explanation regarding these decisions, and I believe that is a major challenge in supervision,” said Tobias Adrian, a senior official at the IMF.
AI is also expected to widen the gap between richer and poorer companies and nations. Efforts to protect against dangerous threats will become increasingly expensive, and companies that actually have good prospects may struggle to protect themselves.
“If we consider the worst-case scenario, attacks often target the weakest points,” said Adrian.
Sarah Breeden, Deputy Governor of the Bank of England, suggested that one potential solution might be the creation of specific insurance schemes, which she compared to deposit insurance used to anticipate bank failures.
“In a cyber context, do we need a system that allows one institution to take over the essential functions of another institution during a disruption?” she asked.
However, the greatest risk is that the excessive success of AI could fundamentally shake the foundations of the global economy. If AI truly fulfils the most optimistic efficiency expectations, machines could replace human roles en masse, leading to widespread unemployment. This would subsequently reduce disposable income and drive the economy towards a recession, thereby weakening the entire system.