Anthropic’s Mythos AI Sparks Cybersecurity Scramble Across U.S. Banks

America’s largest financial institutions are rushing to repair cybersecurity weaknesses uncovered by Anthropic’s advanced AI tool, Mythos, as concerns grow over the banking sector’s aging technology infrastructure.

The powerful AI system, known as Claude Mythos Preview, is being used by several major banks to identify hidden software vulnerabilities across enterprise systems, cloud environments, and proprietary code. According to multiple reports, the technology is exposing weaknesses at a pace traditional cybersecurity teams struggle to match.

Banks, including JPMorgan Chase, Goldman Sachs, Citigroup, Bank of America, and Morgan Stanley, are reportedly among the organizations with access to Mythos through Anthropic’s Project Glasswing initiative.

AI Detects Complex Security Risks Faster Than Humans

Cybersecurity experts say Mythos represents a major shift in how vulnerabilities are discovered and prioritized. Unlike traditional tools that often flag isolated issues, Mythos can connect multiple low-risk flaws into a larger high-risk threat scenario.

Nitin Seth described the development as a “wake-up call” for the banking industry, warning that cyber threats are now moving at machine speed while many security operations still depend heavily on human response times.

The AI model is reportedly uncovering hundreds to thousands of vulnerabilities, many categorized as low or moderate risk but capable of becoming serious threats when combined. Banks are now being forced to patch systems in days rather than weeks.

Aging Banking Technology Under Pressure

One of the biggest concerns raised by Mythos is the condition of legacy banking infrastructure. Experts say the AI is especially effective at identifying weaknesses in outdated software and unsupported systems still widely used across the financial sector.

This accelerated pace of vulnerability detection is creating operational pressure for banks, with some experts warning that institutions may need to temporarily take systems offline more frequently to complete security upgrades.

Smaller Banks Face Cost and Access Challenges

While major Wall Street institutions have gained access to Anthropic Mythos AI, smaller regional banks face significant barriers. The AI model reportedly requires extensive computing resources and carries high operating costs.

Anthropic prices Mythos usage based on token consumption, charging substantially more than its broader Claude AI offerings. To encourage adoption, the company has pledged $100 million in credits for Glasswing partners and select customers.

Larger banks are now sharing findings and threat intelligence with smaller institutions that lack direct access to the platform, helping them prepare for emerging security risks.

AI-Driven Cybersecurity Becomes the New Normal

Cybersecurity leaders increasingly believe AI-powered testing and vulnerability analysis will become standard practice across the financial industry.

Adam Meyers said his team spent days developing entirely new methodologies to effectively use Mythos after gaining access to the tool.

As financial institutions continue integrating AI into cybersecurity operations, experts warn that the race between attackers and defenders is accelerating rapidly, potentially reshaping how banks manage digital security for years to come.