IMF warns new AI models could increase cyber risk for banks

In short: The IMF says newer AI models could make cyber attacks on banks and payment systems more likely and more disruptive.

What happened

The International Monetary Fund, or IMF, warned that the latest AI models could raise cyber risk to a level that could hurt the wider financial system. It said policymakers should prepare for an “inevitable” breach of financial institutions’ cyber defences.

In a blog post published Thursday, IMF officials said advanced AI can cut the time and cost needed to find and exploit security weaknesses. A “vulnerability” is a flaw in software that can be used like an unlocked door. The IMF said this could lead to “correlated failures”, meaning many organisations could be hit around the same time because they use the same popular software and shared service providers.

The warning comes as regulators watch tools like Anthropic’s Claude Mythos. Anthropic said last month Mythos found thousands of high-severity vulnerabilities, including in major operating systems and web browsers. Anthropic is rolling Mythos out gradually, starting with about 40 mostly US-based organisations, including Amazon, Microsoft, and large banks such as JPMorgan Chase, so they can fix issues with “patches” (small updates that close security holes).

Why it matters

If multiple banks or payment networks are disrupted at once, everyday services can be affected. That can include card payments, bank transfers, and access to cash. The IMF warned that confidence could drop quickly if people think several institutions are having problems at the same time.

The IMF also said cyber risk crosses borders, and countries with fewer resources may be more exposed. It urged more international cooperation, along with cyber stress tests and better planning so systems can recover fast after an attack.

Source: Financial Times