JAPAN Law and Practice Contributed by: Hiromi Hayashi, Daisuke Tsuta, Masaki Yukawa and Keiichi Bando, Mori Hamada & Matsumoto
In June 2023, the PPC issued a guidance to users and providers of generative AI. Users were advised to not input personal data to generative AI unless the use of AI is within the purposes of use notified to them and there is assurance that the input data will not be used for machine learning. The guidance suggested that providers not collect sensitive data, with such collection requiring data subjects’ consent, and disclose the purposes of use to data subjects. Copyrights Consideration In the process of machine learning, copyrighted works may be copied or adapted, which may cause copyright infringements. To promote AI developments, the Copyright Act grants an exemption allowing the use of copyrighted works to the necessary extent without permis - sion from the copyright owner where the use (i) does not aim to let the user or others enjoy thoughts or sentiments expressed in the work, and (ii) does not unjustifiably harm the copyright owner’s interests, taking into consideration the type and purpose of the work and the manner in which the work is used (Copyright Act, Article 30-4). The exemption covers the use of copyrighted works for extracting informative elements from a large amount of copyrighted or other works and analysis. The Copyright Act was amended in 2019 to clarify that the exemption covers not only statistical analysis but also deep learning, and not only copying but also transmission for grid computing. In light of the recent increased use of genera - tive AI, copyright holders have concerns about free riding. To responding such concerns, in July 2024, the Agency for Cultural Affairs issued a report called “General Understanding of AI and Copyright in Japan”. It examines, among oth -
ers, whether using copyrighted works in each phase of (a) AI development and learning and (b) generating content by AI and utilisation of AI- generated content may cause copyright infringe - ments. For example, if AI learning is conducted for the purpose of generating a content which is similar to copyrighted works learned by AI, it would not meet requirement (i) of the foregoing exemption and thus such using would constitute an infringement. The report also states that the Agency for Cultural Affairs will continue to collect new information such as further development of AI technology and evolving situations in other jurisdictions and may update the report. Protection of “Shared Data with Limited Access” To promote data sharing between businesses so that big data will be more widely used, the amendment to the Unfair Competition Preven - tion Act introduced the protection of “shared data with limited access” which is defined as any technical or business information: • that is accumulated in a reasonable amount by electronic means; • that is provided to specified persons as a business; and • the access of which is controlled by elec - tronic means. Information controlled as a secret was previ - ously excluded from protection even if the fore - going conditions were met, but under a 2023 amendment, from April 2024, that information also became protected as long as it is not pro - tected as a trade secret. Most typically, where data such as location of smartphones or cars is collected by a business operator (data holder) and sold to third-party business operators for a fee (or shared among business operators in a consortium without charge) under the condi -
175 CHAMBERS.COM
Powered by FlippingBook