Discussing the impact of reviving and utilizing deceased persons’ voices through AI tools, and the sufficiency of the legal frameworks in India and other countries to address this phenomena, we are pleased to bring to you this guest post by Julia Anna Joseph and Snehal Khemka. Julia is an Associate at Cyril Amarchand Mangaldas, Bengaluru and Snehal is an Associate at Chadha and Chadha IP, Gurugram. Views expressed here are those of the authors’ alone.
Voice Clones and Legal Tones: The Intersection of Artificial Intelligence and Posthumous Personality Rights
By Julia Anna Joseph and Snehal Khemka
Generative Artificial Intelligence (“AI”), a game-changing phenomenon in modern day life creates art, plans daily tasks, analyzes data, generates music, and much more. The rise of a multifunctional tool that nearly substitutes a human being, has raised several legal and ethical questions. Recently, a concerning trend has emerged involving the use of generative AI to produce songs mimicking the voice of deceased singers (see here and here). Reviving popular artists posthumously, though nostalgic, prompts critical questions about the boundaries around consent, personality rights, moral rights, economics of the industry, and even ethics. The question is- when AI starts hitting the high notes, who holds the rights to the encore?
Indian Legal Landscape
In India, artists are performers who have performer’s rights under Section 38A and 38B of the Copyright Act, 1957. However, this is not apt to suit the rights of persons whose voices are replicated through AI since a ‘performer’ under the Act is a person who makes a performance, and their rights extend only to the visual or acoustic presentation which they have performed. This is why when voices are cloned through AI, only personality rights can be claimed.
Personality rights empower individuals to safeguard their persona and regulate the commercial usage of their identity, encompassing their voice, likeness, name, image, and other unique identifying characteristics. The justification for approaching the issue of why the present conundrum should be protected from the perspective of personality rights is three fold- First, the justification under privacy rights; Second, the justification to prevent unauthorized commercial use of the artist’s identity; Third, the ethical reason to restrain such uses.
In the realm of AI-generated content, this raises significant questions regarding the extent of protection afforded to an individual’s voice, particularly as AI becomes increasingly adept at replicating vocal attributes convincingly.
In India, the regulations governing AI and personality rights intersection are still nascent. Presently, India has no specific statute that acknowledges the right of personality. Nonetheless, Indian courts, through a series of judicial precedents have begun safeguarding voice as a part of personality rights, illuminated through the intersection of privacy and publicity rights.
Judicial Precedents on Personality Rights
The focus in India initially, was on defining personality rights in relation to privacy rights. The landmark judgment passed in R. Rajagopal v. State of Tamil Nadu (Auto Shankar case) is perhaps the most renowned decision recognizing personality rights vis a vis privacy rights (albeit not explicitly using the term “personality right”) in India, where the Supreme Court upheld an individual’s legitimate right and interest to control the commercial use of their identity. The Court ruled that private matters cannot be published without consent unless derived from public records, emphasizing a balance between freedom of the press and the right to privacy. The focus then shifted to protecting attributes of public figures such as iconic appearances and dialogues, and preventing unauthorized profiting from their fame, as seen in the 2014 Rajinikanth case.
More recently, courts have specifically addressed the unauthorized use of a celebrity’s voice. In 2023, the Delhi High Court issued an omnibus ex-parte injunction prohibiting entities, and the world at large, from leveraging AI to profit off actor Anil Kapoor’s name, likeness, or voice (including the word “Jhakaas”) without his explicit consent. The ruling firmly solidified protecting individuals’ personality rights against any activities aimed at defaming or endangering their image. Similarly, in 2024, the Delhi High Court, in Jackie Shroff’s case declared the unauthorized use of his personality attributes by an AI chatbot platform as an infringement of his personality rights (including the usage of the word “Bhidu”). This ruling holds particular significance as it marks one of the first instances where a court has restrained an unlicensed chatbot created using AI from mimicking a celebrity.
However, the aspect of personality rights recognized in all the aforementioned rulings pertain exclusively to living persons, and the extension of such rights to the dead is still a gray area.
Echoes of the Past: Posthumous Consent and Vocal Resurrections
In India, personality rights lack posthumous application or inheritance. Consequently, an individual’s reputation, personality rights, and associated publicity rights terminate upon their death and are not transferable to their heirs. This stance raises pertinent questions regarding the validity of posthumous consent obtained passively from the deceased singers’ families. While families may grant initial consent, the indefinite creation of new content could lead to disputes over future use, compensation, and preservation of the artist’s legacy. Furthermore, treating a cloned voice as a property could introduce complex legal issues, such as its division under succession laws and the potential commercial interests behind familial consent, especially in cases of strained relationships. Additionally, ethical considerations emerge concerning singers who may have desired their voices to R.I.P (literally), highlighting the unintended consequences of voice revival.
Ethical Considerations
The UNESCO Recommendation on the Ethics of Artificial Intelligence and The Principles for the Ethical Use of AI in the UN System (“Principles”), emphasize the relevance of human rights, dignity, and equality in the use of AI. The Principles outline that States must create enabling environments for media to exercise ethical use of AI systems in their tasks. While the term ‘ethics’ has subjective understandings, in the present context of AI, the ethical value of primacy for respect and dignity of the dead is presumed. Moreover, the Principles emphasize caution in AI systems, specifically with regards to having a ‘defined purpose, necessity and proportionality.’ The term ‘necessity’ here is crucial, considering that one of the immediate questions that pop in one’s mind in this regard is whether the creation of AI-generated music using the voice of a deceased person is necessary at all, considering that art has continually evolved with new artists and fresh talent.
The law must consider ethics, as ethical standards guide human conscience. The two popular approaches to normative ethics are deontological and teleological theories.
Deontological ethics looks at only the motives behind an action to evaluate ethics. In Kantian deontological ethics, Kant professes that actions must arise from a duty and that humanity should be treated as an end in itself and not just a means to an end. This entails respecting each individual’s dignity and autonomy which would forbid the use of a demised person’s characteristics for commercial gain sans consent. Kant also set forth that one’s external actions must be such that the free exercise of one’s will can coexist with freedom of all others. Teleological ethics on the other hand assesses the outcomes of actions.
Re-generating deceased artists’ voices clashes with emerging artists’ ability to pursue music careers since voice cloning can be a cheaper and more accessible alternative. This potentially acts as an entry barrier for emerging singers and also disincentivizes hiring singers, thus contravening ethical principles under both deontological and teleological perspectives. .
(One) of the Ways Forward
As clear from the above discussion, it is clear that unfortunately the law is not able to keep up with the rapid technological advancements. Looking at our counterparts, while no specific regulation globally addresses the recreation of deceased singers’ voices in commercial art, there is growing discussion in favor of having a legislation to govern concerns arising out of replicating attributes of individuals (living and dead) via AI.
However, having a separate legislation to protect the rights of deceased artists against the possibility of exploitation via AI comes with a separate set of issues which includes- determining what exactly shall be covered under it, criteria for a person to be deemed a “celebrity” to fall under the purview of this legislation, adopting and justifying suitable measures to access and store the attributes of personalities of these celebrities, etc. Considering this unique crossroads of the need of protecting personality rights against the AI use on one hand and the challenges with enacting a law to do so on the other, one low hanging fruit (solution) here could be to extend limited posthumous protection of celebrity rights, to their families for a limited period.
As said above, by virtue of the Sushant Singh Rajput case, present Indian jurisprudence holds that personality rights die with a person. However, the economic benefits that arise out of commercial use of one’s persona, the ethical concerns in allowing such unfettered use give scope for its inheritance by successors. In case of the deceased, it also protects the dignity attached to such a person and their family, assuming that inheritors would act in the desired interests of the deceased. Additionally, the Court in the Sushant Singh Rajput case had not taken into consideration the issue of replicating someone’s attributes via AI and thus, holds a very limited applicability in the present context.
Is there scope for recognising it as property law and then under inheritance? At first glance at least, this seems unlikely. However, perhaps in combination with some AI regulations, and keeping copyright’s term limit in mind, something could be developed? In any case, this line of thinking would require far greater examination of the complex legal issues involved.
Although not very common, certain nations worldwide recognize celebrity rights posthumously. For instance, the US state of California, under the California Civil Code, Section 3344.1, extends personality rights for 70 years post death. The Indiana Code Title 32, Section 36-1 provides publicity rights for a person’s lifetime and an additional hundred years after death. Several judgments in the US have also recognized these rights, in Factors etc. Inc. v. Pro Arts Inc.and Martin Luther King v. American Heritage Products Inc. There also exist practices whereby artists transfer their right of publicity to a trust who can take action against unauthorized commercial exploitation of the deceased person’s personality. In Germany, Section 23(3) of the Law Regarding the Author’s Rights provides that people shall be entitled to image rights from birth until 10 years after death, a posthumous right over one’s persona reflected solely in an image.
As AI evolves, balancing artists’ freedom with respect for individual rights and legacy will be a critical challenge. Despite judicial recognition of personality rights, the absence of a legislative framework leaves significant gaps, particularly concerning posthumous personality rights. Though the above suggestion is merely an example of how this issue can be addressed from an IP perspective, it is also crucial that an interdisciplinary approach is adopted, encapsulating perspectives from other areas like information technology law and criminal law.