Japan will hold a meeting involving the country's three megabanks and the Bank of Japan governor to check potential risks posed by Anthropic's artificial intelligence model Mythos.

CoinPost, a blockchain outlet, reported on Thursday that Finance Minister Satsuki Katayama (가타야마 사쓰키) plans to meet executives from Mitsubishi UFJ Financial Group, Sumitomo Mitsui Financial Group and Mizuho Financial Group to share awareness of related risks.

Attendance by BOJ Governor Kazuo Ueda (우에다 가즈오), Japanese Bankers Association Chairman Katsuhiko Kato (가토 가쓰히코) and Japan Exchange Group Chief Executive Hiromi Yamaji (야마지 히로미) is also being coordinated. The Financial Services Agency proposed the meeting.

Mythos is a next-generation AI model that Anthropic announced on April 7. It has been reported to have capabilities high enough to identify and exploit vulnerabilities in major operating systems and web browsers. Anthropic's security team said Mythos found a zero-day vulnerability in OpenBSD that had existed for 27 years and a video library flaw that had not been detected by typical automated testing. Anthropic is limiting who can access it due to concerns it could be used for cyberattacks.

Concern over Mythos is spreading across the U.S. and Japanese financial sectors. In the United States, Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell summoned Wall Street executives, including from Goldman Sachs and Morgan Stanley, to the Treasury Department last week to discuss responses to systemic risk. Japan's ruling Liberal Democratic Party also held an emergency meeting on April 20, including with the National Cybersecurity Strategy Headquarters, and called for a coordinated response by the Cabinet Secretariat and the Financial Services Agency.

Japan's large banks are also raising their level of vigilance. Mitsubishi UFJ Bank President Masakazu Osawa (오사와 마사카즈) last week identified cybersecurity risk as one of the core risks for financial institutions. Anthropic has established a cybersecurity organisation called Project Glasswing, involving Microsoft and Apple, and is pushing to build defence systems. Overseas megabanks, including JPMorgan Chase, have also reportedly begun non-public testing of Mythos.

A key issue at the meeting is expected to be what response direction Japan's Financial Services Agency and the three megabanks will set out for AI models like Mythos. Attention is also on whether access to Mythos, currently provided on a limited basis, will expand to financial institutions and whether Japanese and U.S. authorities will flesh out coordination on responses to AI-based cyber risks.