Gov't, banks given access to latest Anthropic AI model for security

TOKYO — Japan's government and some banks have been given access to U.S. startup Anthropic's new Claude Mythos artificial intelligence model with advanced cybersecurity capabilities, the government said Wednesday, as concerns mount about cyberattacks led by sophisticated AI.

"There is no doubt that this will contribute to strengthening security" at a time when "it is unacceptable" for Japanese financial institutions to fall behind other countries in this field, Finance Minister Satsuki Katayama told reporters.

Hisashi Matsumoto, minister in charge of cybersecurity, also hailed the granting of access to the government and domestic financial institutions, vowing to take steps to "strengthen the cyber resilience of Japan as a whole."

"It is a significant step forward that the door opened," Matsumoto said, referring to the expansion of access rights granted by Anthropic to like-minded countries, including the Japanese government.

The development comes after Katayama said last week that major Japanese banks have been given access to ChatGPT chatbot developer OpenAI's newest GPT-5.5-Cyber model, which is viewed as having similar advanced cybersecurity capabilities to those of Claude Mythos.

Leading Japanese banks, such as MUFG Bank, were expected to have been allowed to use the Anthropic model, according to a source close to the matter.

The latest granting of access to Anthropic's model came after U.S. Treasury Secretary Scott Bessent said it would come within two weeks when he made a three-day visit to Japan through May 13, according to Katayama.

The Claude Mythos model is seen as highly capable of finding security vulnerabilities, and Anthropic initially limited access to IT companies such as Google LLC and certain financial institutions after its April unveiling, due to concerns it could be misused for cyberattacks.

© KYODO