Japan's National Cyber Director Yoichi Iida told Kyodo News in an interview published 10 August that using sophisticated AI in the country's incoming cyber-defence regime is "inevitable" — and disclosed a dependency failure the government has already lived through.

The June disruption

Iida said the June 2026 suspension of access to Mythos-class models affected security vulnerability scanning of key Japanese government systems. His conclusion was structural rather than diplomatic: Japan should not depend on any single AI model. The claim rests on his account alone — no US government document confirming the suspension or its scope has been produced, and the wire copy does not cleanly separate a US export-control action from the vendor's own access restrictions.

The regime it feeds into

The Active Cyber Defence legislation passed the Diet on 16 May, and Prime Minister Sanae Takaichi's Cabinet set the pre-emptive measures to begin on 1 October 2026. Iida said the government is weighing how far to automate operations while keeping a human in the final decision loop.

The timing argument

His case for automation is the compression of the defender's window: the gap between a vulnerability being discovered and exploited could shrink to "a fraction of a hundredth or a thousandth" of what it once was. At that speed, he argues, human-paced response stops being a policy choice.

An interview, not a decision

No rule, budget or procurement was announced. The 1 October date belongs to the regime as a whole, not to any AI component of it, and nothing here says Japan has adopted autonomous cyber operations.