A Japanese government panel approved a draft plan on Tuesday that asks generative AI companies to reveal more about how they train their systems.
The plan is called a "principle code." It is not a strict law, but a set of guidelines that firms are encouraged to follow.
The draft was presented during an online meeting of an expert panel focused on intellectual property rights in the age of AI.
Why the Code Was Created
Officials have grown concerned that AI companies may be using text and images to train their models without getting permission first.
This raises questions about whether copyright laws are being broken as generative AI tools become more common in daily life and business.
The new code is built on an AI-related law that Japan passed in May 2025. It tries to balance the growth of AI technology with protection for creators and rights holders.
The government plans to use what is called a "comply or explain" system. Under this approach, companies are not forced to follow the rules by law.
Instead, they can choose to follow the code or publicly explain why they have decided not to.
If a company chooses to comply, it must announce this decision on its own website. It must also notify the government of its choice.
The rule will not just apply to Japanese companies. Foreign businesses that offer AI services inside Japan will also be expected to follow the same code.
What the Code Requires
The draft code is built around three main principles.
The first principle asks companies to share which AI models they use. They must also explain what data they used for training and how that data was collected.
This information should be posted publicly on the company's website. However, businesses will not be required to share sensitive information as part of this disclosure.
The second principle deals with copyright disputes. If a rights holder believes their work was used without permission, they can ask a company whether a specific webpage was part of its training data.
The third principle focuses on user protection. It says companies must respond when users raise concerns about possible copyright infringement tied to the AI service they are using.
Together, the three principles are meant to give creators and rights holders more insight into how their work may be used by AI systems.
The plan does not include penalties for companies that choose not to comply, as long as they explain their reasoning publicly.
The proposal now moves forward as part of Japan's broader effort to manage the rise of generative AI while supporting continued innovation in the sector.
