Citi CEO Fraser calls Anthropic’s Mythos ‘not a good day’ for global cybersecurity

When Anthropic released its Mythos AI model to approved users in April, the ripple effects were immediate and far-reaching. The model, capable of identifying and exploiting zero-day vulnerabilities in every major operating system and web browser, has since triggered a cascade of government responses spanning Washington, New Delhi, and Brussels — forcing regulators, financial institutions, and the White House itself into an unprecedented scramble to shore up digital defenses.anthropic

A "Tsunami" of Patching

Speaking at the Qatar Economic Forum's UNGA Special Edition in New York on Sunday, Citigroup ↗ 0.51% CEO Jane Fraser described the fallout in stark terms. "Every company globally is doing massive patching right now, creating a defense race," she said, calling Mythos's release "not a good day". Fraser noted that after Mythos emerged, leaders from the U.S. Treasury Department and the Federal Reserve convened Wall Street executives to discuss the risks and urged firms to fortify their systems.tradingkey+1

The alarm had begun months earlier. According to a detailed account published Sunday by Business Insider and POLITICO, Microsoft ↘ 0.99% had privately warned Trump administration officials in late March that Mythos "represented a significant leap and could automate sophisticated cyberattacks against financial institutions, the energy grid and other critical infrastructure, allowing less skilled actors to carry out operations once limited to elite hackers". Treasury Secretary Scott Bessent subsequently held an emergency meeting with the heads of Bank of America ↘ 0.68%, Goldman Sachs The Goldman Sachs Group, Inc. ↘ 0.36%, Morgan Stanley ↘ 0.55%, Wells Fargo ↘ 1.01%, and Citigroup on April 7, laying out a worst-case scenario in which a layperson could crack the security of banks, power grids, and government systems.businessinsider

Global Regulatory Response

The alarm quickly went international. In May, India's Securities and Exchange Board established a dedicated task force called cyber-suraksha.ai, explicitly naming Claude Mythos as a cybersecurity threat and ordering every regulated entity in the world's fourth-largest equity market to overhaul its cyber defenses. The task force mandated immediate system patching, vulnerability assessments, strengthened API security, and onboarding to a market-wide Security Operations Centre for real-time threat detection.newsonair+2

In Europe, the EU's cybersecurity agency Enisa secured access to Mythos in September after months of negotiations with Anthropic, alongside access to OpenAI's GPT-5.6 Cyber model. Anthropic has also expanded its Project Glasswing initiative, providing early model access to roughly 40 organizations including Google Alphabet Inc. ↗ 0.78%, Apple ↘ 0.52%, Amazon Amazon.com, Inc. ↘ 0.35%, and JPMorgan Chase ↘ 0.84% to test the model as a defensive tool.radware+1

The White House Reckoning

The broader political fallout has been equally dramatic. The Trump administration, which entered office promising a hands-off approach to AI, was forced into its first federal regulatory framework after the Mythos crisis. An executive order signed June 2 established voluntary safety-review guidelines for frontier models, though the process was nearly derailed by last-minute lobbying from former AI czar David Sacks. A subsequent model from Anthropic, called Fable, was taken offline for 19 days in June after Amazon researchers discovered a jailbreak vulnerability, prompting the Commerce Department to invoke export controls.businessinsider

The episode exposed the tension at the heart of American AI policy. "This administration is used to just setting the rules and walking away, but with Anthropic, the power of their technology has kept them in a bilateral conversation with the White House, where very few tech companies could survive," said Matt Calkins, CEO of Appian ↘ 4.50%.businessinsider