Japan Megabanks Seek Access to Mythos AI Model

Tokyo, May 13 (Jiji Press)–Japan’s three megabanks are working to gain access to Claude Mythos, a cutting-edge artificial intelligence model developed by U.S. startup Anthropic, aiming to use it to enhance cybersecurity measures, it was learned Wednesday. On Tuesday, officials of MUFG Bank, Sumitomo Mitsui Banking Corp. and Mizuho Bank met with U.S. Treasury Secretary Scott Bessent, who was visiting Japan. During the meeting, the U.S. side signaled that it would positively consider granting the banks access to Mythos. Anthropic has not publicly released the AI model, and currently, only a few companies and groups have access to it. The model is highly adept at identifying system vulnerabilities, sparking concerns about its possible use in cyberattacks. The Japanese government has also called on the United States to grant the banks access to Mythos. “Although we have yet to obtain access to Mythos, we are doing our best to diagnose system vulnerabilitis,” Toru Nakashima, president of Sumitomo Mitsui Financial Group Inc., the parent of Sumitomo Mitsui Banking, said at a press conference on Wednesday. On Thursday, the Financial Services Agency will establish a working group comprising financial institutions, including the three banks, as well as technology companies, to discuss stronger measures against risks posed by high-performance AI models, such as Mythos. END [Copyright The Jiji Press, Ltd.]