Finance ministers and major central banks worldwide are concerned that the "Claude Mythos" risks shaking the stability of the financial system.

Finance ministers and major central banks worldwide are concerned that the AI ​​"Claude Mythos" poses a risk to financial system security, after discovering its potential to find and exploit cyber vulnerabilities in critical systems, even though it is not yet publicly available.

On April 17, 2026, at 3:05 AM, the BBC reported that: Finance ministers, central bankers, and financial experts worldwide are expressing serious concerns about new artificial intelligence models that could impact the stability of the financial system. Following the development of the Claude Mythos model by Anthropic, it is seen as having unprecedented potential for discovering and exploiting cybersecurity vulnerabilities.

These concerns led to urgent discussions at the international level, including a meeting of the International Monetary Fund in Washington, D.C. François-Philippe Champagne, Canada's Minister of Finance. specify that This issue is serious enough to warrant attention from finance ministers worldwide, with a comparison that while geopolitical risks like the Strait of Hormuz can be defined, the risks from AI are something entirely unknown.

An AI model that's too powerful, so we're hesitant to use it yet.

Claude Mythos is one of the latest models within the Anthropic AI system, developed to investigate AI behavior in situations that might conflict with human values. โดย Developer specify that This model has outstanding capabilities in the field of cybersecurity (computer security).

Despite its ability to find vulnerabilities in IT systems, software, and infrastructure, the company has decided not to release it publicly for fear of it being misused.

However, Anthropic has made its model available for testing by major technology companies such as Amazon Web Services, Microsoft, Nvidia, and CrowdStrike under "Project Glasswing," which aims to enhance the security of critical software worldwide.

High potential, but still controversial.

Despite widespread concerns, some experts note that the model's capabilities still need further testing.

A report from the UK AI Security Institute, the only independent organization that conducted initial testing. Found that Claude Mythos can indeed find and exploit vulnerabilities in weakly protected systems, but it doesn't significantly surpass its predecessor, Claude Opus.

researcher specify that This type of model is likely to be further developed in the future and may become an important tool for both defending and attacking systems.

Banks and the government are beginning to prepare for the situation.

The financial sector began to move rapidly, by... Barclays' senior executives specify that This issue is serious enough that all parties must take it seriously and it is necessary to urgently understand any vulnerabilities that may be exposed.

side Andrew Bailey, Governor of the Bank of England Warning that Such technology could make detecting vulnerabilities in IT systems easier, but on the other hand, it could also open up opportunities for cybercriminals to exploit them in attacks.

The U.S. Treasury Department has also begun discussions with major banks to test the system in advance before the model is widely implemented.

The new era of finance: Opportunities and risks.

Investment expert James Wise. I think that Claude Mythos may be just the beginning of a new generation of AI capable of revealing system vulnerabilities deeper and faster.

Meanwhile, businesses and investors are increasingly interested in companies developing AI-powered security technologies, believing that in the future, models capable of discovering vulnerabilities may be the same tools used to fix them.

refer : www.bbc.com

Read news related to Technology circles around the world can be found here.