Japan's Intellectual Property Strategy Headquarters published the Principle Code on the Protection of Intellectual Property and Transparency for the Appropriate Use of Generative AI on 25 August 2026. The text was settled after a public comment exercise on an earlier draft, and the Cabinet Office Intellectual Property Strategy Promotion Bureau announced the notification start date and the notification form on 8 September 2026. The Code is a voluntary instrument rather than legislation, and it carries no penalty.
Section 1(4) of the Code governs acceptance. A generative AI business that accepts the Code publishes a statement of acceptance on its corporate website, sets out how it has implemented each Principle with reasoned explanations for anything it has not implemented, and files notice with the Cabinet Office Intellectual Property Strategy Promotion Bureau in the prescribed form. Principle 1 covers public disclosure of model information, training data information and accountability measures. Principle 2 lets a copyright holder in litigation ask whether URLs it identifies appear in the training data. Principle 3 lets a user ask whether the training data contains content identical or similar to the output they generated. The Bureau publishes a list of participating businesses with links to their disclosures, and the whole instrument operates on a comply or explain basis.
The Code reaches generative AI developers and generative AI providers that have made systems or services available to the public, and it reaches operators outside Japan whose services are directed at Japanese users. Contractors handling only part of a development, systems built for a single entity or a limited designated group, and unpublished research fall outside it. Model developers running large scale web crawling, application vendors building on third party models, and Japanese subsidiaries of foreign AI companies each have to decide whether to file a notice, and what their crawling and data sourcing disclosures will say.
Notification under Section 1(4) opens on 26 October 2026. Businesses that accept the Code review their disclosures annually and update them. Because the instrument is voluntary and carries no sanction, the practical pressure comes from the published participant list and from the comparison it invites between a company's stated position and its actual data sourcing. A business that declines to implement a Principle is expected to give reasons rather than stay silent.
Licentium advises AI developers, model providers and their investors on disclosure duties across jurisdictions. Work we undertake includes mapping training data disclosure obligations, preparing acceptance statements and notification filings, and reviewing crawling and licensing practice against published commitments.