JPMorgan Chase CEO Jamie Dimon warned that cybersecurity risks posed by Anthropic's Mythos AI model are significant and being closely monitored by the US government. Dimon described giving wide access to Mythos as akin to "giving ballistic missiles to individuals," stressing the danger of its broad release [1, 2, 3].
Anthropic launched the Mythos AI model in April 2026, initially releasing it to a limited number of users including JPMorgan. The model is among the most advanced AI tools designed to detect cybersecurity vulnerabilities and autonomously find zero-day exploits [1, 3, 4].
Due to national security concerns, the US government imposed export controls in June 2026 restricting Mythos 5 and Fable 5 access for foreign nationals. Anthropic responded by temporarily cutting off global access. After the company implemented new safeguards, the Department of Commerce lifted these restrictions on June 30, allowing access restoration [1, 3].
Anthropic CEO Dario Amodei estimated it could take 6 to 12 months to fix the thousands of vulnerabilities identified by Mythos [5]. Meanwhile, financial institutions such as JPMorgan and Bank of America have limited Mythos use to internal security testing and are sharing vulnerability findings with peers [4].
Brian Moynihan, CEO of Bank of America, said that tools like Mythos will drastically increase the volume and speed of cybersecurity defense work banks must undertake, along with the urgency needed to patch exploited weaknesses [4]. The model’s ability to both detect and potentially expose software flaws has raised concerns among financial regulators and national security agencies [4].
Crypto exchanges including Coinbase and Binance are also exploring Mythos to enhance their security testing, according to sources [5].
