Japan eyes cutting-edge AI under pre-emptive cyber defence policy

TOKYO: Japan is considering employing advanced artificial intelligence (AI) systems as part of its preemptive cyber defence policy to counter the risk of attacks from sophisticated models like Anthropic's Claude Mythos, Kyodo News reported. The National Cyber director, Yoichi Iida, stated that Japan will undoubtedly utilize cutting-edge AI technology due to growing concerns over AI's ability to detect security vulnerabilities.

Iida predicted that the time between identifying a vulnerability and its exploitation could be reduced to a fraction of a hundredth or thousandth of what it previously was.

The Claude Mythos model, known for its high capability in finding security vulnerabilities, was initially restricted to IT companies and specific financial institutions in April, as Anthropic feared potential misuse for cyberattacks. This follows Prime Minister Sanae Takaichi's Cabinet decision in March to allow the government to take active cyber defence measures from October 1 to prevent severe cyberattacks that could disrupt critical infrastructure, economic activity, and people's livelihoods.

Iida emphasized that while Japan needs to consider how to advance automation in its cyber defence operations, human involvement must be maintained in the final decision-making process. He did not elaborate on concrete steps but cautioned against relying on a single AI model, citing the US government's export control directive in June which suspended access to Mythos-class models.

This suspension impacted security vulnerability scanning operations for key government systems. Iida added that if advanced AI models were used in nation-sponsored cyberattacks, they could become significantly more sophisticated, with the intensity of such offensives expected to rise.