How should Voices be Legally Protected in the Age of AI?

人工智慧時代下語音之法律上保障

目前日本關於語音保護之法律架構被認為缺乏實效性，特別是在涉及個別名人聲音，及濫用普通人聲音進行詐欺和勒索等犯罪行為之情況。基於商業用途模仿名人聲音之法律責任的不確定性，使該問題更加複雜。雖然人工智慧技術將普通人的聲音用於犯罪目的，但未經許可錄製對話卻是被允許的。於此情形，實有必要探究日本以及國際上是如何在法律上保護語音。本文深入探討了日本的語音保護法，隨後分析美國之法律現況，並強調公開權及禁止未經授權錄音之相關法律。本文提出保護名人聲音之四個標準：(1)該聲音廣為人知；(2)該聲音被用於商業目的；(3)聲音被有意模仿；(4)模仿聲音之相似度足夠使消費者誤以為是該名人之真實聲音。至於普通人聲音之保護，應禁止企業經營者未經許可錄製對話，以防止建立語音資料庫。惟在勒索、詐欺或騷擾案件中，受害者應有權利錄製與潛在犯罪人之對話，以作為證據。

The current legal framework for voice protection in Japan is deemed ineffective, particularly in cases involving distinctive celebrity voices and the misuse of ordinary voices for criminal activities such as fraud and extorsion. The uncertainty regarding legal liability for imitating celebrities in commercial uses adds complexity. While AI technology exploits ordinary voices for criminal purposes, recording conversations without permission is permitted. This situation necessitates urgent research on how voices are legally protected in Japan and globally. The paper delves into Japan’s voice protection laws, followed by an analysis of the United States’legal landscape, emphasizing rights of publicity and laws against unauthorized recording. The author proposes four criteria for protecting celebrity voices: (1) the voice is widely known, (2) the voice is used for a commercial purpose, (3) the voice is imitated intentionally, and (4) the imitations are similar enough for consumers to mistake them for the person’s actual voice. To safeguard ordinary voices, recording conversations without permission by business operators should be prohibited to prevent voice database creation. However, victims should retain the right to record conversations with potential criminals as evidence in cases of extortion, fraud, or harassment.