Banks Face New AI‑Powered Cyber Threats

A fresh wave of artificial intelligence tools is set to reshape the way cyber attackers approach banks. The latest model, named Mythos, can write code and act on its own, giving it the power to spot hidden weaknesses in old and new computer systems alike. Experts warn that such a tool could help hackers craft attacks faster than security teams can respond. The model’s coding skill lets it scan complex networks that mix modern software with legacy programs built decades ago. Banks, which rely on a handful of common platforms for customer checks and transactions, are especially vulnerable. When Mythos finds a flaw, it can quickly design an exploit that could damage many institutions at once. Government officials in the United States, Canada and Britain have already met with top bank leaders to discuss this risk. The Treasury has urged financial firms to prepare for a wave of market changes that could include AI‑driven attacks. While the creators of Mythos say it will not be released to the public, they are inviting a select group of tech giants and security vendors to test it and build defenses.Security researchers have already used the model to uncover dozens of serious vulnerabilities in popular operating systems and browsers. One example is a 16‑year‑old flaw in an audio‑video library that was still active. When a model like Mythos can find such hidden bugs, it raises the stakes for institutions that run older technology stacks. The banking sector is urged to rethink its security posture from the ground up. Open‑source collaboration could help spread knowledge of how to defend against AI‑generated threats, but many firms remain silent on the issue. The growing conversation signals that banks must act quickly to protect their customers and infrastructure from a new generation of cyber attackers.