Particle.news

Japan Revises AI Basic Plan to Treat High‑Performance Models as National Security Risk

The draft mandates continuous reviews, inspections, alerts to critical infrastructure, coordination with foreign governments, support for AI‑generated content detection, and stepped‑up defenses

Overview

  • The government published a draft revision to its AI Basic Plan on Friday that labels attacks enabled by high‑performance AI as a security risk and says strengthening defenses is indispensable.
  • The revision responds to new capability‑rich models such as Anthropic’s Claude Mutus, which can find software flaws quickly and lower the skill barrier for attackers.
  • The plan calls for active, continuous reviews of systems, targeted investigations and inspections, and early warnings to critical infrastructure operators to speed detection and response.
  • Tokyo also proposes deeper international coordination with foreign governments and AI developers and will support tools to detect AI‑generated content to counter false or misleading information.
  • Industry voices say greater automation in defensive tools is the practical way to keep pace with AI misuse, and the government will open the draft for public comment before seeking near‑term cabinet approval.