Japan Seeks Access to Anthropic AI, Worried that Claude Mythos Will Be Used for Cyber Attacks

TOKYO - The Japanese government is negotiating with Anthropic, an American artificial intelligence company, to gain access to the latest AI model, Claude Mythos. Tokyo wants to use the technology to strengthen cyber defenses amid concerns that AI-based attacks are becoming more sophisticated.

Quoted from Kyodo News, several government sources said Claude Mythos was said to be very capable of finding security gaps in web browsers and other software.

This ability is a serious problem. On the one hand, AI can help countries close security holes. On the other hand, information about the gap can be misused to attack various systems.

Japanese Prime Minister Sanae Takaichi has asked Cybersecurity Minister Hisashi Matsumoto to prepare measures to deal with cyber attacks using the latest AI.

Japan also wants to gain knowledge and expertise in AI development. According to a government source, access to Claude Mythos is considered important because it is expected that advanced AI systems will continue to emerge.

Japanese and Anthropic officials are expected to hold talks soon. Japan has also begun preparing a technical environment to use Mythos if access is granted.

Matsumoto said Japan was exchanging views with the United States on the negotiations.

"At this stage when we do not have access to the model, we still have to continue the countermeasures," said Matsumoto after the cabinet meeting.

Currently, access to Mythos is still limited. The model introduced by Anthropic in April can only be used by technology companies such as Google LLC and a number of financial institutions. However, governments and public organizations are also beginning to apply for access.

Experts are concerned that Mythos could be misused to disrupt systems in many sectors. AI-based cyber attacks are no longer just about websites or applications, but can also affect financial services, transportation, public services, and critical infrastructure.

Anthropic is known as a developer of generative AI that emphasizes safety aspects. The company was founded in 2021 by former OpenAI executive Dario Amodei, and released the Claude conversational AI in 2023.

The issue will also be discussed at the G7 forum. The finance ministers and central bank governors of the developed countries are expected to discuss the risks of the latest AI-based cyber attacks, including Mythos, at a meeting in Paris next week.

In Japan, Chief Cabinet Secretary Minoru Kihara stressed that the government would move quickly to respond to possible abuse. The Japanese Financial Services Agency has also established a public-private framework to strengthen countermeasures.