Japan wins access to Anthropic's Claude Mythos for cyberdefense
Claude Mythos already found thousands of high-severity vulnerabilities, now targeting Japan's financial systems.
On May 22, 2026, Japan's Finance Minister Satsuki Katayama announced that the Japanese government will be granted access to Anthropic's latest AI model, Claude Mythos, later in May. The move follows a three-day visit by US Treasury Secretary Scott Bessent through May 13, during which he promised access to the Japanese government and financial institutions within two weeks. According to sources, Japan's three megabanks—MUFG Bank, Sumitomo Mitsui Banking Corp., and Mizuho Bank—are expected to gain access to the sophisticated AI model, which was released by Anthropic in April 2026.
Claude Mythos is currently restricted to IT companies like Google and select financial institutions due to safety and security concerns. Anthropic reports that its preview has already identified thousands of high-severity vulnerabilities. The access is part of a broader US-Japan pledge to respond quickly to cyberattacks targeting vulnerabilities in financial systems. On the same day, Japan's Financial Services Agency urged financial institutions to implement short-term countermeasures, including suspending system operations, to mitigate heightened risks from advanced AI-driven cyber threats.
- Japan's government and three megabanks (MUFG, Sumitomo Mitsui, Mizuho) get access to Claude Mythos by late May 2026.
- Claude Mythos preview has already identified thousands of high-severity vulnerabilities.
- Access follows US Treasury Secretary Bessent's promise during a May 13 visit, part of US-Japan cyberdefense collaboration.
Why It Matters
Giving Japan financial institutions Claude Mythos access could dramatically improve real-time cyberattack detection and response.