Introduction
The movie Her, released in 2013, tells a romantic story about a lonely man who falls in love with an AI assistant with a charming voice. Absurd as it may seem, deepfake technology has made such scenario possible.
Deepfakes, a recent buzzword, are an eye-catching application of Artificial Intelligence (AI). The term deepfake refers to an “AI-generated or manipulated image, audio or video content that resembles existing persons, objects, places, entities or events and would falsely appear to a person to be authentic or truthful”.[1] For example, the former voice assistant in ChatGPT-4 was considered to sound “eerily similar” to the famous American actress Scarlett Johansson.[2]
Deepfake technology increases “the speed, precision, and scale” of voice and face cloning to an unprecedented level.[3] As early as 2018, the world’s first AI news reporter made its debut at the World Internet Conference, which impersonated a real journalist of Xinhua News.[4] As this technology advances, deepfakes can now, in some occasions, convincingly deceive human eyes and ears.
Consequently, new business opportunities have emerged, particularly in the fields of art, entertainment, and media.[5] In the movie industry, deepfake technology can empower actors and directors to overcome obstacles like aging, voice injuries, and even mortality. For example, the character Luke Skywalker appears significantly younger in the Mandalorian compared to his actor, Mark Hamill, thanks to this technology.[6] In the entertainment industry, deepfakes featuring celebrities’ visual likenesses or voices have become part of global culture. For instance, the deepfake track Heart on My Sleeve, which some argue sounds like a collaboration between artists Drake and The Weekend, has gone viral on many platforms.[7]
Despite its technical and economic benefits, deepfakes also pose considerable risks to individuals, industries, and society as a whole. Harmful deepfakes present multiple challenges to a broad range of areas, including privacy, consumer welfare, authenticity of information, democracy, and national security.[8] Among these, this paper focuses on protecting individual-level interests against harmful deepfakes. This protection is important for economic and moral reasons. Deepfakes can be market substitutes for real performers, thereby driving them out of business. Meanwhile, they offer cost-efficient solutions to amplify reach in, e.g., marketing, training, and content creation. They may also impact individuals’ non-economic interests, such as privacy and reputation.
Lawsuits against unauthorized deepfakes featuring performers’ voices or their appearance (likenesses) are on the rise globally. Recently, the Beijing Internet Court issued a landmark ruling on AI developers’ liability for creating and distributing deepfakes.[9] Similarly, Lovo, an AI voice generator in the U.S., has been sued by two voice actors for imitating their voices without authorization.[10]
In response to the above-mentioned challenges, the U.S. Copyright Office (the Office) proposed a new digital replica right.[11] This proposal is based on the concern that current U.S. laws are not sufficient to protect individuals’ voices or likenesses used in deepfakes.[12] At the same time, across the Atlantic, only a few studies have shed light on the EU’s regulatory framework that governs the making and sharing of deepfakes.[13]
Against this background, this paper explores to what extent individuals can control and profit from deepfakes featuring their voices or likenesses in the EU. Section II introduces the background of the proposed digital replica right in the U.S. Sections III and IV examine the extent to which the EU regulatory framework protects individuals’ moral and economic interests in deepfakes that impersonate them. Section V concludes with some proposals to ensure that individuals’ rights are sufficiently protected in the EU.
* Zizhuo Zhou is an IPR Researcher at 4iP Council. Any opinions expressed in this article are those of the author and do not necessarily represent the views of 4iP Council and of any of its supporters.
[1] Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 on harmonised rules on artificial intelligence, amending Regulations (EC) No 300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1139 and (EU) 2019/2144, and Directives 2014/90/EU, (EU) 2016/797, (EU) 2020/1828 [2024] OJ L1689/12 (Artificial Intelligence Act), art. 3(60).
[2] Matt Murphy and Liv McMahon, ‘Scarlett Johansson “Shocked” by AI Chatbot Imitation’ BBC (21 May 2024) <https://www.bbc.com/news/articles/cm559l5g529o> accessed 4 November 2024.
[3] U.S. Copyright Office, ‘Copyright and Artificial Intelligence Part 1: Digital Replicas’ (Copyright Office, July 2024) iii <https://www.copyright.gov/ai/Copyright-and-Artificial-Intelligence-Part-1-Digital-Replicas-Report.pdf> (The U.S. Copyright Office’s Report).
[4] ‘World’s First AI News Anchor Unveiled in China’ The Guardian (London, 9 November 2018) <https://www.theguardian.com/world/2018/nov/09/worlds-first-ai-news-anchor-unveiled-in-china> accessed 4 November 2024.
[5] Vejay Lalla and others, ‘Artificial Intelligence: Deepfakes in the Entertainment Industry’ WIPO Magazine <https://www.wipo.int/wipo_magazine/en/2022/02/article_0003.html> accessed 4 November 2024;
[6] ‘The Mandalorian: Luke Skywalker Deepfake Technology “Could be Used in Really Bad Ways”’ BBC (London, 29 August 2021) <https://www.bbc.co.uk/newsround/58340104> accessed 4 November 2024.
[7] Joe Coscarelli, ‘An A.I. Hit of Fake “Drake” and “The Weeknd” Rattles the Music World’ New York Times (New York, April 19, 2023) <https://www.nytimes.com/2023/04/19/arts/music/ai-drake-the-weeknd-fake.html> accessed 4 November 2024.
[8] Chesney, Bobby and Danielle Citron. ‘Deepfakes: A looming challenge for privacy, democracy, and national security.’ (2019) 107 Calif. L. Rev. 1753.
[9] Cao Yin, ‘AI-generated Voice Mimicry Infringes on Rights’ China Daily (Beijing, 24 April 2024) https://www.chinadaily.com.cn/a/202404/24/WS66286719a31082fc043c3acf.html accessed 4 November 2024.
[10] Cade Metz, ‘What Do You Do When A.I. Takes Your Voice?’ The New York Times (New York, 16 May 2024) https://www.nytimes.com/2024/05/16/technology/ai-voice-clone-lawsuit.html accessed 4 November 2024.
[11] The U.S. Copyright Office’s Report 29.
[12] ibid.
[13] See Huijstee Mariëtte van and others, Tackling Deepfakes in European Policy (Publications Office of the European Union, 2021) 39; Fabian Rößling, ‘Regulierung von Deepfakes’ (2024) ZfDR 187; Murat Karaboga, ‘Die Regulierung von Deepfakes auf EU-Ebene: Überblick eines Flickenteppichs und Einordnung des Digital Services Act- und KI-Regulierungsvorschlags’ in Sylvia Jaki and Stefan Steiger (eds), Digitale Hate Speech (J.B.Metzler 2023).
