Particle.news

Japan’s Panel Approves Non‑Binding AI Principle Code Requiring Training‑Data Disclosure

The guidance seeks to boost transparency to curb copyright and privacy risks from generative AI and to shape industry behavior without new laws.

Overview

  • The government presented a draft “Principle Code” to its AI intellectual property review panel on Tuesday and the panel gave broad approval with formal adoption expected soon.
  • The code centers on three disclosure duties: a public overview of training data, a process for responding to rights‑holder requests, and information obligations toward AI users.
  • It targets system developers and service providers offering AI to Japan, including overseas firms, and asks companies that accept the code to publish acceptance on their websites and notify the government.
  • Although the guidance carries no legal force, reports say it would use a comply‑or‑explain regime and may include penalties for operators who intentionally acquire personal data or act in bad faith.
  • The code builds on last May’s AI law and relies on public reporting and market pressure to protect creators; its real effect will depend on firms’ voluntary uptake and any future enforcement steps.