JPMorgan CEO Warns Against Broad Access to Anthropic's AI Model Mythos

Jamie Dimon, CEO of JPMorgan, has issued a strong warning about providing wide access to Anthropic's AI model Mythos, comparing it to "giving people ballistic missiles." Anthropic unveiled Mythos in April but initially restricted public access due to its powerful security vulnerability detection capabilities and concerns over potential cyberattack misuse. The company first allowed select organizations to use the model and in June launched a limited version called Pible, while expanding Mythos access to more organizations. Shortly after, the Trump administration ordered a block on access to both models but reversed the decision about a week and a half later.

Dimon praised the Trump administration's intervention, stating at a Pennsylvania conference that the government's scrutiny was necessary to ensure control over the model upon release. He emphasized that Mythos represents a serious issue because it effectively arms users with highly dangerous capabilities. In a CNBC interview, Dimon also discussed corporate spending on AI models, highlighting the rapid rise in costs and the need for companies to monitor employee usage carefully. He noted that businesses are becoming more savvy about managing AI expenses, negotiating with suppliers, and optimizing queries to use the cheapest tokens and data centers, a trend he expects to intensify.