“We are excited about what responsible AI can do for music,” Koda’s chief Gorm Arildsen said. “But innovation can’t be built on stolen goods.” (source: techxplore)
Recently, the Bombay High Court has granted landmark interim protection to legendary playback singer Asha Bhosle against AI voice cloning. Justice Arif S. Doctor remarked that “making AI tools available to enable the conversion of any voice into that of a celebrity without his/her permission would constitute a violation of the celebrity’s personality rights.”
As tech companies train generative AI models on publicly available celebrity content, the core question is whether the‘fair dealing’ exception under copyright law can shield AI developers from liability when they scrape voices, images, and distinctive mannerisms of famous personalities without consent?
“Courts in India treat personality rights as distinct from copyright, and even publicly available voice recordings cannot be freely used for AI training without consent,” answered Zameer Nathani, global general counsel at DNEG. “In the US, White vs Samsung Electronics (9th Circuit Court) held that imitating a celebrity’s likeness violates publicity rights, while Midler vs Ford Motor Co. (9th Circuit Court) the court held that imitation of a singer’s voice is actionable,” cited Nathani.
Delhi High Court, in recent times, has passed several injunctions protecting Kumar Sanu, Aishwarya Rai Bachchan, Abhishek Bachchan, and journalist Sudhir Chaudhary against AI-generated deepfakes underscoring underscoring a growing judicial recognition that voice, image, likeness, and mannerisms constitute protected personality attributes regardless of public availability or fair dealing claims.
“Recent court orders passed in India protecting personality rights are mostly ad interim orders,” noted Ashutoh Kumar, senior partner at Singh and Singh. “It is yet to be seen how a fair dealing defence would be applied in relation to these rights. However, judicial trends indicate a strong inclination towards protecting individual identity and preventing unauthorised commercial exploitation of one’s personality right.”
Personal data made publicly available
Compounding the uncertainty is India’s yet-to-be-enforced Digital Personal Data Protection Act, 2023. Section 3(c)(ii) exempts from the Act’s protections “personal data made publicly available by the data principal”
“The exemption applies only to personal data, not to broader personality attributes,” voiced Kumar. “Unauthorised commercial use of such personality traits may constitute misappropriation or passing off, regardless of whether the underlying data was publicly accessible or not.”
The Supreme Court’s 2017 decision in Justice K.S. Puttaswamy v. Union of India established privacy as a fundamental right, creating a constitutional floor for personality protection that statutory exemptions cannot circumvent.
“The DPDP Act exempts data “made publicly available by the data principal,” but courts are unlikely to extend this to AI scraping that repurposes data without consent.” noted Nathani while referring to cases including Aishwarya Rai Bachchan vs Unauthorised Websites (Delhi High Court, 2025) where an interim relief was granted against AI-generated sensitive material using her likeness.
Two-tier Analysis
The ANI Media v. OpenAI case, sub-judice before the Delhi High Court, represents India’s first significant test of whether AI training amounts to copyright infringement. An Amicus Curiae in the case has proposed a two-step framework:: first, determine if the use falls under statutory exceptions such as private use, criticism/review, or reporting of current events; second, conduct a fairness analysis to evaluate the use in question.
India currently protects personality rights through Article 21 of the Constitution, common law doctrines like passing off, and moral rights under Sections 38A-38B of the Copyright Act.. Post the decision on this case, Indian courts may adopt the proposed two-step analysis.
“The courts in India are likely to adopt a dual-tier approach—first applying fair dealing analysis, then separately assessing personality misappropriation,” said Nathani
“Availability of fair dealing defence and sufficiency of the same against a claim of personality rights violations, is a debatable one as often the personality rights assertions are based on breach of privacy claims or on law of passing off,” commented Kumar.
Legislation or Litigation
In October 2025, Denmark proposed an amendment to its copyright law granting individuals ownership rights over their face, body, and voice to counter the misuse of AI-generated deepfakes.
As India navigates these new-age tensions through judicial interpretations, several legislative pathways emerge:
First, Parliament could enact dedicated personality rights legislation codifying protections for voice, image, gesture, and distinctive style, with explicit provisions addressing AI training mechanisms.
Second, the Ministry of Electronics and Information Technology could clarify through subordinate DPDP Act rules that the Section 3(c)(ii) exemption does not extend to biometric or personality-identifying attributes used for AI training absent explicit consent.
Lastly, the Copyright Act could be amended to exclude personality attributes from Section 52 fair dealing or introduce a statutory licensing regime for AI training akin to Section 31D broadcasting provisions.
“For India a privacy-by-design and human rights in terms of data, privacy, intellectual property rights including personality rights aligned protocols are crucial for AI compliance in India’s evolving legal jurisprudence,” concludes Nathani
This will be discussed at our IP and Innovation Conclave on November 14 in Mumbai. Check out the agenda and register now.
‘ The preceding article may include information circulated by third parties ’
‘ Some details of this article were extracted from the following source legal.economictimes.indiatimes.com ’
