Generative AI using "voices" and "likenesses" without authorization: Japan's METI proposes regulatory plan, with "imitation" and "coincidental similarity" not included in regulation
生成AIによる「声」「肖像」の無断利用、経産省が規制案…「まね」「偶然似ている」は対象から除外
On the 13th, Japan's Ministry of Economy, Trade and Industry submitted a specific proposal to a panel of experts, planning to add to the Unfair Competition Prevention Act regulations on generative AI using others' voices and likenesses without authorization; imitation and coincidental similarity are excluded from the scope of regulation. The proposal lists three categories of regulated conduct: using a likeness or voice itself as a product for appreciation, using it for the purpose of product differentiation, and using it in advertising—for example, photo cards made with a person's own photo, products adorned with a likeness, and voice data made with a voice actor's voice, all aimed at commercially exploiting the appeal of likenesses and voices. At the expert meeting, some raised that social media use for profit should also be included in regulation. Excluding imitation and voice imitation, and the use by a person who coincidentally resembles someone of their own likeness or voice, is intended to clarify the scope of regulation and curb the impact on legitimate expressive activities. Based on the criticism that such unauthorized use is difficult to regulate adequately under the current law, METI had previously studied amending the law, and will continue discussions at the expert meeting and proceed with concrete institutional design.