US Banks Begin to Change, Preparing for Major Disaster

US banks are beginning to improve their security systems, particularly weaknesses identified by Anthropic’s AI tool, Mythos, which was launched recently. Following its launch, a small number of banks have gained access to Mythos, ultimately allowing the company to uncover issues within them. Several sources also indicate that large banks are assisting smaller ones without access to prepare their systems, as cited from Reuters on Wednesday (13/5/2026). In addition, Wall Street banks testing Mythos found that the model connects low-risk vulnerabilities into high-risk ones. This discovery enables them to check whether software has been updated or not. A source from one major bank explained that Mythos can identify vulnerabilities in proprietary code and open-source. Therefore, banks must upgrade legacy technology where software support has reached end-of-life. Meanwhile, another source explained that Mythos can reveal hundreds to thousands of low-to-medium rated vulnerabilities that must be fixed. According to that source, Mythos requires banks to repair vulnerabilities quickly, even within days, compared to previously waiting much longer. Early access to Mythos was previously given to several companies in the Glasswing initiative and an additional 40 organisations, including JPMorgan Chase, Goldman Sachs, Citigroup, Bank of America, and Morgan Stanley. One of those with access is Crowdstrike, a cybersecurity company that is part of Project Glasswing. Crowdstrike’s leader, Adam Meyers, said they spent several days learning how to use the system before discovering security gaps. “We spent a full weekend figuring out the best way to use the system before starting to look for security gaps,” he explained.