JPMorgan Chase’s chief executive sounded the alarm Wednesday about the risks tied to Anthropic’s Mythos AI model, calling them a “real issue” while noting that the federal government is already paying close attention to the matter.
Speaking at Senator Dave McCormick’s Pennsylvania Defense and Innovation Summit, the JPMorgan CEO drew a stark comparison to illustrate the danger, stating that access to advanced AI tools must be tightly controlled. “You’re giving ballistic missiles to individuals with Mythos,” he warned.
Anthropic, a U.S.-based artificial intelligence company, first made Mythos available in April to a limited group of users that included JPMorgan. Financial institutions have taken particular interest in the model because it is regarded as one of the most capable AI systems when it comes to detecting cybersecurity weaknesses, allowing companies to find and address vulnerabilities at a much faster pace.
Despite that appeal, the U.S. government stepped in during June and ordered Anthropic to cut off foreign nationals from accessing its most advanced AI models — Fable 5 and Mythos 5 — citing national security concerns. That directive forced the company to pull worldwide access to those systems.
Access was later restored after Anthropic put new protective safeguards into place to the government’s satisfaction.
The situation reflects a broader struggle playing out between governments and technology companies over how to handle AI safety, particularly when it comes to systems capable of identifying software flaws. Federal officials have increased their scrutiny of new AI model launches, driven by fears that adversarial nations — including China, Russia, and others — could exploit cutting-edge AI for military or intelligence purposes.
Anthropic was not available to respond to the JPMorgan executive’s comments, as they were made outside of normal business hours.
