Japan’s government agencies and financial institutions are among around 150 organizations in over 15 countries that have been granted access to Anthropic’s cutting-edge artificial intelligence model, Claude Mythos, for purposes of cybersecurity testing and research. This initiative requires participating organizations to meet stringent security standards before they can utilize the model to uncover system vulnerabilities and bolster defenses against potential cyber threats.
Japanese Finance Minister Satsuki Katayama acknowledged the inclusion of Japanese government institutions and several banks in this program, calling it a crucial step in preparing for future cybersecurity challenges. Claude Mythos is categorized as “frontier AI,” a term used for highly advanced systems capable of identifying weaknesses in complex digital infrastructures. While these capabilities are invaluable for security research, experts caution that they could be exploited to target critical infrastructure if not properly managed.
In addition to Japan, the European Union has also gained access to this technology, with the European Union Agency for Cybersecurity (ENISA) set to take part in the program. European officials have expressed support for the initiative, viewing it as part of ongoing efforts to enhance cyber resilience across the continent.
Meanwhile, OpenAI is reportedly in the process of offering one of its own advanced AI models to major Japanese financial institutions, highlighting the increasing global interest in employing frontier AI to fortify cybersecurity defenses. This move underscores the international momentum toward leveraging advanced AI technologies to address the growing complexities of digital security.