Japan’s Largest Banks to Access Anthropic’s Claude Mythos Model this Month
In Focus
- Anthropic will onboard the Japan’s megabanks on Mythos by the end of May
- U.S. Treasury Secretary informed the banks about the access during a meeting in Tokyo
- Anthropic has limited Mythos access to about 40 U.S. partners
- The AI firm plans to expand access to banks in the U.K. and the EU
Three of Japan’s largest banks will start accessing Anthropic’s Claude Mythos Preview in about two weeks, sources familiar with the issue said. This will be the first time that megabanks in Japan will be accessing Mythos. Anthropic has limited Mythos access to selected American partners.
When Will Japanese Banks Gain Access to Mythos?
The three financial institutions, Mitsubishi UFJ Financial Group, Mizuho Financial Group, and Sumitomo Mitsui Financial Group will be onboarded onto Anthropic’s banking AI model by the end of this month.
The banks were informed about the decision during a meeting between the U.S. Treasury Secretary Scott Bessent and Japan’s Finance Minister Satsuki Katayama in Tokyo. Following the meeting, Japan announced plans to set up a a public-private working group to address the cybersecurity risks that Mythos poses to the Japanese financial system.
Mythos rollout in Japan represents an enterprise win for Anthropic in Asia’s financial services sector. It also positions the company as a viable alternative to OpenAI within the banking sector. For Japanese banks, access to Mythos accelerated AI integration across workflows in the banking industry.
Claude Mythos Preview debuted on April 7, 2026 as part of Anthropic’s cybersecurity initiative, Project Glasswing. The model is designed to identify security flaws and vulnerabilities within software.
What are the Cybersecurity Concerns Around Claude Mythos?
Cybersecurity experts have said that Mythos presents significant hacking risks and advised caution when interpreting vulnerability results generated by the AI system. This advice has prompted action by policymakers and regulators, who have issued warnings about the AI model.
Last month, regulators across Europe, Australia, and South Korea started assessing the banking risks posed by the AI model amidst growing concerns about its potential impact on financial cybersecurity. Anthropic has not made Mythos public yet. Instead, the AI firm is running a controlled rollout with a group of 40 launch partners.
They include Apple, Google, AWS, NVIDIA, Microsoft, and Palo Alto Networks. Several American banks, including JPMorganChase have also been given access to the AI Model. Anthropic, which recently introduced new agentic tools for financial institutions, plans to expand Mythos access to banks in the U.K. and the European Union.
What Does Mythos Access Mean for Japan?
The working group established by Japanese Finance Minister Satsuki has been tasked with identifying exposures in Mythos, executing defensive measures, and developing contingency plans that would guide the rollout of a patching process across the country’s financial system.
Anthropic and OpenAI teams in Japan are part of the working group. Despite this, the banks involved will have to adhere to Anthropic’s terms on output disclosure. Mythos is used to identify vulnerabilities in partner systems and generate remedies for the same. Anthropic prohibits partners from publishing exploits.
