TOKYO - The Japanese government held an emergency meeting with the Bank of Japan and executives from the country's three megabanks to discuss potential risks posed by the new artificial intelligence model 'Claude Mythos' to Japan's financial system.
BOJ Governor Kazuo Ueda and senior banking leaders gathered on the afternoon of April 24 for the hastily arranged talks with the Financial Services Agency.
The meeting reflected growing concern over the rapid advance of AI technology and its possible misuse in cyberattacks.
Financial Services Minister Satsuki Katayama said, "The Takaiichi administration aims to build a strong economy, and we must prevail in this battle over AI as well. This is truly a crisis already upon us."
At the center of the concern is 'Claude Mythos,' a newly announced AI model unveiled on April 7. The word 'Mythos' comes from ancient Greek and means 'myth.'
Experts say the model's most striking strength lies in its programming ability.
AI researcher Shota Imai said the system is capable of discovering software bugs that had gone unnoticed for 15 to 25 years.
That capability means it could also quickly identify vulnerabilities hidden in operating systems and web browsers, making it a potential tool for cybercriminals if misused.
Analysts warn that attacks on medical systems or financial networks could have severe consequences.
"If such power were directed at healthcare systems that safeguard lives or financial systems that manage money, the damage would be enormous," Imai said. "It could threaten social infrastructure and even national security."
In a worst-case scenario, experts say risks could include tampering with deposit systems, fraudulent transfers and theft of account information, with major consequences for everyday life.
The day 'Mythos' was announced, U.S. Treasury Secretary Scott Bessent reportedly convened executives from major banks for emergency talks and began strengthening security measures.
The model was developed by U.S.-based Anthropic, founded in 2021 by former OpenAI executives.
Anthropic rose to prominence through its safety-focused generative AI model 'Claude' and has expanded rapidly.
In October 2025, Chief Executive Dario Amodei visited the prime minister's office and exchanged views directly with Prime Minister Sanae Takaichi.
More recently, the company has reportedly clashed with the Trump administration over military uses of AI. 'Claude' is believed to have been used in attacks on Iran.
Amodei has warned of three major dangers: autonomous AI behavior, misuse by individuals or governments, and the possibility of economic disruption.
Citing concerns over the model's powerful capabilities, Anthropic has postponed a full public release of 'Mythos.'
Instead, it has limited access to around 50 companies and organizations, including major technology firms such as Google and Microsoft, while launching a project aimed at strengthening security.
Katayama said, "New preparations are needed against the fires of conflict that AI progress could bring to the financial sector."
As governments around the world move quickly to respond, Japan has also established a working group, with its first meeting expected soon.
Source: TBS
