Threat of New AI Model: Ensure Financial System Is Fully Protected

Editorial

16:07 JST, May 17, 2026

There have been growing concerns worldwide that Claude Mythos, a new artificial intelligence model developed by U.S. startup Anthropic, could threaten the stability of financial systems.

The government should work with the private sector to urgently check for any vulnerabilities in the domestic financial system and expedite the development of defenses against cyberattacks.

Anthropic released Mythos on April 7. Meant to help companies strengthen their security against cyberattacks, this AI model was originally created to check for system vulnerabilities.

Mythos reportedly identified, in a short period of time, a vulnerability lurking within a system renowned as one of the world’s most robust. Experts had not been able to find it for 27 years. Conversely, this incident has heightened fears that unscrupulous people could exploit this AI model to launch cyberattacks.

In light of this, Anthropic has limited the availability of Mythos to major U.S. tech giants such as Google and Microsoft, out of consideration for the potential risks associated with the release of this AI to the public.

Which side — attackers or defenders — will gain the upper hand? This issue can be said to symbolize a new era in which a battle over AI is unfolding.

The financial sector, in particular, is increasingly concerned about Mythos. The financial system is a foundation of people’s lives. If hit by a cyberattack, it could suffer catastrophic damage.

Specifically, concerns have been raised over risks such as major leaks of personal information and the possibility of triggering a domino effect of failures across the entire financial system due to panic selling of massive amounts of assets.

The Financial Services Agency has set up a public-private working group, and its first meeting has been held. To enhance their security, Japan’s three major megabanks have reportedly gained access to Mythos.

Outdated systems that have been in place for many years are still used for domestic payment systems. Given the identified vulnerabilities, efforts must be accelerated to first identify system-related issues. It is also crucial to confirm the steps for responding to a crisis involving market infrastructure and other critical matters.

The competition in AI development has been fierce. In April, the U.S. company OpenAI also announced a new model with enhanced capabilities for detecting system vulnerabilities. There is strong speculation that Chinese companies will also successfully develop high-performance AI models in the not-too-distant future.

It is essential to strengthen international cooperation to determine how to deal with these issues globally. Finance ministers and central bank governors from the Group of Seven countries are scheduled to meet in Paris soon. They should discuss in detail how to strengthen regulations to prevent the global financial system from faltering.

(From The Yomiuri Shimbun, May 17, 2026)