The AI model spooking the world’s biggest banks

Anthropic's new AI model Claude Mythos is causing concern across the global banking sector due to its advanced cybersecurity offensive capabilities. A select group of US banks including JPMorgan, Goldman Sachs, Morgan Stanley, Bank of America, and Citigroup have been granted access through Anthropic's Project Glasswing initiative, while European banks remain without access. Regulators at the IMF spring meeting in Washington flagged cybersecurity risks, and the British government's AI Security Institute assessed Mythos as substantially more capable at cyber offence than any previously evaluated model. Banks without access have raised fairness concerns, and European supervisors are monitoring the situation through existing cyber resilience frameworks.