BlockBeats News, June 4th, Japan will introduce Anthropic's Claude Mythos for bank network defense, with major banks such as Mitsubishi UFJ, Resona, and Sumitomo Mitsui gaining access.

Japanese Finance Minister Kozue Katayama stated that the government has established a public-private joint working group to assess the risks this model may pose to the financial system. Anthropic's introduction states that the unreleased Claude Mythos Preview is designed for defensive cybersecurity work, capable of local vulnerability detection and penetration testing. It has identified thousands of high-risk vulnerabilities in major operating systems and browsers, including multiple zero-day vulnerabilities.

Katayama also revealed that some Japanese financial institutions had previously obtained network security support from OpenAI GPT-5.5 through restricted access arrangements.

This move indicates that companies like OpenAI and Anthropic are advancing the application of advanced code models in the financial industry's network security field, positioning them as crucial tools to enhance the defense capabilities of key infrastructure.

---------------------------------
Click on the original text link below to join the DCS Beating · Feishu AI News Channel, which monitors global AI hotspots and news 24/7.