US and European regulators step up oversight of banks’ AI risks — Cyprus Mail
Banking regulators in the United States and Europe are increasing their focus on artificial intelligence risks, including potential cyberattacks on banks and the actions of autonomous AI systems which, according to sources, could destabilize financial markets. This was reported by Cyprus Mail, citing six informed sources.
US regulators, including the Federal Reserve, the Federal Deposit Insurance Corporation and the Office of the Comptroller of the Currency, are asking banks about their use of advanced AI models. They are interested in whether institutions use Chinese-developed technologies, AI agents in cyber defense, and whether they are able to control increasingly autonomous systems.
Control over autonomous systems
US supervisory authorities are also determining what restrictions banks impose on AI agents, how they monitor and approve their actions, and whether they can track and audit the decisions of such systems. They are separately asking about safeguards that prevent agents from exceeding their authority, human involvement in key decisions, and the ability to switch off autonomous systems.
According to the publication’s sources, regulators are currently gathering information and have not set specific mandatory requirements for banks. Written and oral inquiries are being sent to large, medium-sized and small financial institutions.
More current news is available on the UA.News Telegram channel Telegram.
Recovery after cyberattacks
In Europe, the Bank of England and the European Central Bank are urging banks to strengthen plans for restoring operations after possible AI-related disruptions. British supervisory authorities recommend that the largest banks assess recovery plans based on independent technical infrastructure that can be deployed if primary systems are compromised.
In July, the ECB set banks a deadline at the end of October to prepare plans to counter AI-enabled cyber threats, including improving backup and recovery. Eurozone banks are also being asked to identify exactly where AI is used in their operations and by third-party providers in order to understand critical technological dependencies and be able to shut down and restart systems in a controlled manner.
Grant Thornton partner Michael Woodbridge said that AI implementation often outpaces the creation of proper control mechanisms. BCG representative Sukand Ramachandran stressed that for protection against attacks, rapid recovery is no less important than preventing them.
