Japan AI guidelines urge against intellectual property misuse

The Japanese government urged generative artificial intelligence (AI) developers on Tuesday to prevent the misuse of intellectual property as it released new draft guidelines aimed at protecting creators' rights, Tokyo-based Kyodo News reported.

The draft guidelines establish a framework to balance rapid technological innovation with copyright protection.

While not legally binding, the rules urge AI developers to maintain transparency by publicly disclosing how they collect data to train their models and advise them explicitly advising them to avoid gathering data from pirate websites.

Under the framework, operators would be encouraged, upon request and under specific conditions, to disclose whether copyrighted materials, such as manga, music, and films, were used in their training datasets.

This mechanism is designed to assist copyright holders who discover AI-generated works that closely resemble their creations and need evidence to prepare for potential litigation.

The guidelines also ask developers to set up internal notification systems to inform users if training datasets contain copyrighted works that are similar to the generated content.

The rules, which stem from an AI law enacted last year and are expected to be finalized soon, will apply to both domestic companies and foreign operators providing AI services in Japan.

To avoid excessive corporate burdens or stifling innovation, the government stated that mandatory disclosure will not be required for trade secrets or security-related data.

Although the guidelines carry no legal penalties, non-compliant companies will be asked to explain why they refuse to comply.