Approximately 150 organizations spanning over 15 countries, including Japanese government agencies and financial institutions, have been granted access to Anthropic’s cutting-edge AI model, Claude Mythos, specifically for cybersecurity testing and research purposes. This initiative requires participating organizations to adhere to stringent security protocols before they can utilize the AI to pinpoint system vulnerabilities and bolster defenses against potential cyber threats.
Japanese Finance Minister Satsuki Katayama has confirmed that both government bodies and numerous banks in Japan are part of this initiative. She emphasized that access to Claude Mythos represents a crucial step in preparing for future cybersecurity challenges. This AI model is part of a category known as “frontier AI,” distinguished by its ability to detect weaknesses in sophisticated digital systems. While these models hold significant potential for enhancing security research, experts have underscored the importance of controlling their use to prevent misuse against critical infrastructure.
The European Union has also secured access to Claude Mythos, with the European Union Agency for Cybersecurity (ENISA) anticipated to engage in the program. European officials have expressed approval of this development, considering it a vital component of ongoing efforts to enhance cyber resilience across the region.
In a separate development, OpenAI is reportedly preparing to provide access to one of its own advanced AI models to major Japanese financial institutions. This move underscores the growing international interest in leveraging frontier AI technologies to reinforce cybersecurity defenses globally.