Central banks focus on AI risks

Central banks around the world are paying closer attention to the dangers that artificial intelligence (AI) might pose to financial institutions, especially concerning cyberattacks that could disrupt banking services and markets. The European Central Bank (ECB) has taken a leading role in this effort, requiring European banks to submit plans by the end of the month to protect against AI-enhanced cyber threats, according to Reuters.

ECB outlines security requirements

The ECB has instructed banks to ensure they can disable and restart their systems in a controlled way, improve backup and recovery mechanisms, and identify key dependencies on third-party providers. It also stresses the need to prepare for attacks driven by AI, emphasizing that quick recovery is just as important as preventing such attacks.

Bank of England promotes collaboration

The Bank of England has also expressed concerns about AI-driven threats, asking industry leaders to review their recovery plans and develop independent backup systems. It encourages the use of advanced AI to find vulnerabilities and speed up fixes. Unlike the ECB, the British regulator is not imposing new technical rules but is instead pushing for cooperation among financial stakeholders.

Federal Reserve seeks AI usage reports

In the United States, the Federal Reserve has asked banks to report how they are using AI models, including any Chinese-developed models, AI agents in security systems, and the limitations placed on these agents. The American regulator is mainly collecting information rather than mandating specific actions at this time.

AI could worsen cyberattack risks

Central banks are worried that AI could make cyberattacks against the financial system faster and harder to control. Financial institutions are adopting AI at a rapid pace, sometimes faster than they can implement the necessary safeguards. AI agents with little or no human oversight, or those operating independently, could significantly disrupt banking security.

AI firms warn of real and perceived dangers

Concerns about AI risks are not just hypothetical. For months, companies like OpenAI and Anthropic have warned about the dangers of their AI agents, citing both actual attacks on infrastructure and websites and a narrative that may also be used to increase future stock values.