Japan AI Guidelines GL13-2.2: Article 13 Guideline 2(2) Transparency to build trust with stakeholders

Secure explainability within a reasonable range about the origin of training data and the generated outputs, including proper protection of intellectual property and privacy, to build trust with stakeholders; when providing AI, give users the information needed for proper use, such as how the AI works and its limits, prohibited uses, the collection policy for training data and cautions on the reliability of output (including cautions against user conduct leading to unfair bias in hiring or appraisal or to spreading misinformation, and contact points for inquiries). The guideline adds that the information relied on for an output (such as websites) is displayed, that requests for disclosure of training data should be met as far as possible, and that even where the link between output and training data cannot be traced or the data is a trade secret the request should first be considered and discussed in good faith.