Fed, Treasury Warn Morgan Stanley of Mythos AI’s Thousands of Zero-Day Vulnerabilities
Scott Bessent and Jerome Powell convened Morgan Stanley CEO Ted Pick with four other systemically important bank chiefs to highlight cybersecurity threats from Anthropic’s Mythos AI model, which identified thousands of zero-day vulnerabilities. Banks face pressure to upgrade defenses and increase security spending to mitigate AI-driven attack risks.
1. Urgent Treasury-Fed AI Cybersecurity Briefing
On Tuesday, Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened Morgan Stanley CEO Ted Pick alongside four other systemically important bank leaders at the Treasury building to review cybersecurity risks posed by Anthropic’s new Mythos AI model. The session emphasized the need for immediate industry awareness of potential attack vectors exposed by advanced AI capabilities.
2. Capabilities of Mythos AI Model
Anthropic’s Mythos AI has demonstrated the ability to locate and exploit thousands of zero-day vulnerabilities without specialized security training, prompting the company to restrict access through Project Glasswing. Only a select group of launch partners received usage credits to conduct defensive security testing and assess the model’s impact on critical infrastructure.
3. Implications for Morgan Stanley
Morgan Stanley may need to accelerate its cyber defense initiatives, potentially increasing security budgets and staffing to address AI-driven threats. Heightened regulatory scrutiny and evolving compliance requirements could further impact the firm’s operational planning and cost structures in the coming quarters.