\u3000In light of the current situation surrounding generative AI services, CODA requests the following from businesses engaged in the development and provision of generative AI services:<\/p>\n\n\n\n
\u3000In recent years, generative AI services have been announced one after another, particularly by overseas generative AI businesses. AI technology has significant potential, including improving work efficiency and supporting creative activities.
\u3000However, for AI to be accepted by society with confidence, it is essential to respect the rights of content creators and copyright holders. Even in the case of generative AI services deployed across borders, the rights of Japanese copyrighted works and creators must, naturally, be respected.<\/p>\n\n\n\n
\u3000However, under the current state of generative AI services, serious issues that cannot be overlooked have emerged:<\/p>\n\n\n\n
\u3000Even if generated outputs are not publicly disclosed, this issue cannot simply be dismissed as falling within \u201cprivate use by users.\u201d The core issue is that it is the AI businesses\u2019 services that are generating outputs highly similar to existing copyrighted works, and that such services are being broadly provided to the general public by AI businesses with enormous influence.<\/p>\n\n\n\n
\u3000Regarding AI training, we are aware that there is a view that, based on Article 30-4 of the Japanese Copyright Act, such use may be permitted in certain cases as \u201cuse for non-enjoyment purposes.\u201d However, under the current situation, training materials are being reproduced as concrete outputs, generating images\/videos identical or highly similar to existing Japanese content. We believe that training which results in such generation constitutes \u201cuse for enjoyment purposes,\u201d and even if training may have aspects of non-enjoyment purposes, it effectively coexists with enjoyment purposes in practice.
\u3000Accordingly, we believe that, in many current service operations, the acts of reproduction carried out during the training process themselves may constitute copyright infringement.
\u3000Furthermore, from the perspective of U.S. copyright law, many outputs generated by current generative AI services are difficult to recognize as \u201ctransformative\u201d uses that add new expression or meaning to the original copyrighted works and are substantially different in purpose and character from the use of the original works. Such generation also has the potential to harm the economic value of existing and potential markets for copyrighted works, and therefore we believe such uses do not qualify as fair use. The same applies to training activities that result in such generation.<\/p>\n\n\n\n
\u3000Under copyright law, there is no system whereby infringement liability is avoided simply by responding after a rights holder raises an objection. The use of copyrighted works should be based on prior authorization. If it is technically possible to recognize a connection between generated outputs and existing copyrighted works, then AI businesses have a responsibility to establish mechanisms to prevent problems before they occur, including by implementing filters at the output stage for copyrighted works for which prior authorization has not been obtained.<\/p>\n\n\n\n
\u3000In addition, the Japanese government has also expressed the view that \u201cwith respect to the use of copyrighted works, it is clear under both domestic law and treaties that authorization from rights holders is the principle, and the current situation in which unauthorized use is occurring cannot be overlooked.\u201d<\/p>\n\n\n\n
\u3000Even recently, multiple cases have been reported in which generated outputs created and published using generative AI were criticized due to similarities with existing copyrighted works, despite such similarities not being intended by users. These situations are being recognized as risks associated with the use of generative AI, and as a result are becoming factors that discourage the use of generative AI.<\/p>\n\n\n\n
\u3000Generative AI should be used as a technology that respects and supports human creativity. Generating images\/videos highly similar to the expressions of existing copyrighted works through unauthorized use constitutes a serious problem that undermines creators\u2019 rights and the foundation of creative activity.
\u3000The current situation, in which images\/videos highly similar to existing copyrighted works are being trained on, generated, and output without authorization from rights holders, constitutes copyright infringement. CODA hereby reiterates its position and calls for this understanding to be properly shared.<\/p>\n\n\n\n
(Translated using an automatic translation system.)<\/p>\n\n\n\n
About CODA<\/strong>
CODA (Content Overseas Distribution Association) was established in 2002 at the call of the Ministry of Economy, Trade and Industry and the Agency for Cultural Affairs to promote the overseas development of Japanese content and take anti-piracy measures. Japan\u2019s proud content, including music, movies, anime, broadcast programs, video games, and publishing, plays an important role in enhancing the nation\u2019s international presence and economic growth. As digital technology becomes more widespread, it is even more significant to protect Japan\u2019s content from increasingly artful copyright infringement and promote the content industry\u2019s development. CODA contributes to the deterrence and detection of online and other piracy, by sharing knowledge with relevant government agencies, organizations, and companies in Japan and abroad, to engage in direct and indirect anti-piracy measures, as well as public relations activities. Visit https:\/\/coda-cj.jp\/en\/activity\/<\/a> for more information on CODA\u2019s projects.<\/p>\n","protected":false},"featured_media":0,"parent":0,"template":"","news_cat":[3],"class_list":["post-900","news","type-news","status-publish","hentry","news_cat-info"],"aioseo_notices":[],"_links":{"self":[{"href":"https:\/\/coda-cj.jp\/en\/wp-json\/wp\/v2\/news\/900","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/coda-cj.jp\/en\/wp-json\/wp\/v2\/news"}],"about":[{"href":"https:\/\/coda-cj.jp\/en\/wp-json\/wp\/v2\/types\/news"}],"wp:attachment":[{"href":"https:\/\/coda-cj.jp\/en\/wp-json\/wp\/v2\/media?parent=900"}],"wp:term":[{"taxonomy":"news_cat","embeddable":true,"href":"https:\/\/coda-cj.jp\/en\/wp-json\/wp\/v2\/news_cat?post=900"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}