In approximately two weeks, Japanese megabanks are set to gain access to Anthropic’s advanced AI system, Mythos, which has demonstrated a remarkable ability to identify and chain vulnerabilities in legacy IT systems, thus posing challenges for banking cybersecurity. This development follows the testing of Mythos by select U.S. banks, which has expanded the path for access by financial institutions in Japan and Europe. In response to the potential risks posed by advanced AI models like Mythos, Japan has formed a public-private working group to address and mitigate financial system risks.

Anthropic: Anthropic is an AI research firm specializing in safe, interpretable large language models like its Claude series under constitutional AI frameworks. Its Mythos preview model excels at autonomously detecting and exploiting software vulnerabilities, prompting global cybersecurity reevaluations. The company plans to provide Japan’s megabanks access to Mythos, following grants to U.S. institutions amid regulatory scrutiny.
Scott Bessent: Scott Bessent serves as U.S. Treasury Secretary, overseeing financial regulation and international economic coordination. He has urged banks to bolster defenses against AI-driven cyber threats, including meetings with CEOs on Mythos capabilities. His recent Tokyo discussions with Japan’s Finance Minister supported the pathway for Japanese megabanks to access Anthropic’s Mythos.
Satsuki Katayama: Satsuki Katayama is Japan’s Minister of Finance and Financial Services, guiding policies on economic stability and banking oversight. She has convened bank executives to discuss AI cybersecurity threats from models like Mythos. Following her meeting with U.S. Treasury Secretary Scott Bessent, she announced a public-private working group convening this week on Mythos risks.
Mizuho Financial Group: Mizuho Financial Group operates as a leading Japanese financial institution offering banking, securities, and trust banking services worldwide. It prioritizes regulatory compliance and technological upgrades for financial stability. As one of Japan’s megabanks, it is set to receive access to Anthropic’s Mythos to address cybersecurity challenges posed by the AI model.
Mitsubishi UFJ Financial Group: Mitsubishi UFJ Financial Group is Japan’s largest banking group, providing retail, corporate, and investment banking services through subsidiaries like MUFG Bank on a global scale. It focuses on digital innovation and cybersecurity enhancements amid emerging AI threats. The group is expected to gain access to Anthropic’s Mythos AI model in about two weeks to evaluate its impact on banking systems.
Sumitomo Mitsui Financial Group: Sumitomo Mitsui Financial Group is a prominent Japanese megabank delivering comprehensive financial solutions including consumer finance and global corporate services. It engages actively in fintech advancements and risk mitigation strategies. The group will gain access to Anthropic’s Mythos AI model alongside peers to test vulnerabilities in legacy infrastructure.

`json
{“Access Expansion”: “Japan megabanks are slated to gain access to Mythos, with select U.S. banks having already tested the system.”, “Mythos Capabilities”: “Mythos challenges banking cybersecurity by demonstrating significant capabilities in identifying and chaining vulnerabilities in legacy IT systems.”, “Regulatory Response”: “Japan establishes a public-private working group to address financial system risks associated with advanced AI models like Mythos.”}
`