Japan Downplays Impact of Anthropic Freeze But Warns of Rising Digital Vulnerabilities

By Yang Jie and Megumi Fujikawa

The U.S.'s suspension of advanced artificial-intelligence tools being tested by Japanese banks poses no operational threat as trials remain in the early stages, Japan's finance minister said Tuesday.

Speaking to reporters after the abrupt global halt of Anthropic's Claude Mythos models, Finance Minister Satsuki Katayama downplayed immediate domestic fallout but voiced concern over rising digital vulnerabilities following the models' brief public availability prior to the U.S. intervention.

Katayama said Tokyo has yet to receive an assessment of the security risks from Washington and will press for a full explanation at the G-7 summit on Wednesday.

She said Japanese financial institutions are still operating inside a closed test environment. Given the early stage of testing, she said the freeze has little immediate bearing on Japan's financial system, and that existing channels with U.S. regulators remain in place.

To hedge against further U.S. regulatory moves, she said Japan intends to accelerate its own defensive capabilities.

Katayama pointed to a project involving OpenAI and SoftBank Group to develop models such as a "defense AI" for Japan's domestic market.

Her comments came as SoftBank unveiled plans Tuesday to deploy specialized cybersecurity services in Japan using OpenAI's advanced models.

Offered through a joint venture with the ChatGPT maker, the service will help local entities identify vulnerabilities and plan remediation, with a focus on securing critical infrastructure such as financial institutions, power facilities and transportation tools, SoftBank Chairman and Chief Executive Masayoshi Son said.

News Corp, owner of The Wall Street Journal and Dow Jones Newswires, has a content-licensing partnership with OpenAI.

Write to Yang Jie at jie.yang@wsj.com and Megumi Fujikawa at megumi.fujikawa@wsj.com

(END) Dow Jones Newswires

June 16, 2026 07:17 ET (11:17 GMT)