Artificial intelligence and intellectual property rights collide as Japan steps up pressure on generative AI firms to reveal what powers their models.

The Japanese government has approved a draft “principle code” calling on generative AI companies to disclose information about their training data and the methods used to collect it, as Japan seeks to strengthen intellectual property protection while supporting technological innovation.

The draft code was presented at an online meeting of an expert panel studying intellectual property rights in the AI era. It builds on an AI-related law enacted in May 2025.

The proposed framework would use a nonbinding “comply or explain” approach, under which businesses could follow the code or publicly explain why they choose not to. The framework would also cover foreign businesses providing AI services in Japan.

The principle code is centered on three pillars regarding training data: disclosure of an outline, disclosure to rights holders and disclosure to AI users. Under the first, AI businesses would disclose information about the models they use, their training data and the methods used to collect it.

The second covers requests from copyright and other rights holders about whether specific webpages were included in training data. The third covers information requests from users concerned about potential copyright infringement.

The draft has also drawn attention to the need for continued review of its effectiveness and the treatment of sensitive information, including trade secrets.

The government plans to finalize the principles and begin implementing them as early as this autumn.

Also Read: NYT Publisher: ‘Tech Giants Strip-Mine Original Journalism, Repackage Stolen Goods as Their Own’