INDIA Law and Practice Contributed by: Mohit Goel, Sidhant Goel, Aditya Goel and Mehr Bajaj, Sim and San, Attorneys at Law
Know-how may also be assigned as part of business or asset transfers. There are no statutory formalities requiring recordal, notarisation or registration, but written agreements are strongly recommended for evidentiary and commercial certainty. Cross-bor- der transactions must also comply with the Foreign Exchange Management Act, 1999, applicable tax laws and, where relevant, competition law requirements. 6.6 Reverse Engineering Reverse engineering of lawfully acquired products is not prohibited under Indian law: a person may analyse a product and use the resulting information for com- petitive purposes. The position narrows if: • the product was acquired under a contractual restriction prohibiting reverse engineering; • the subject matter is protected by a registered right; or • the information is used as a springboard from a prior breach of confidence. Contractual prohibitions on reverse engineering are broadly enforceable between contracting parties. The Copyright Act creates a specific exception for reverse engineering of computer programs to achieve interop- erability, under Section 52 (1)(ab). 7. Data, AI and Emerging IP Issues 7.1 Data Rights and Database Protection India does not recognise a sui generis database right; instead, datasets are protected through copyright, contract and trade secret principles. Copyright sub- sists only where a database reflects originality in the selection or arrangement of data, not in the underlying facts. Consequently, businesses rely heavily on licens- ing agreements, confidentiality obligations and trade secret protection to safeguard proprietary datasets. Data scraping is not inherently unlawful but is assessed under copyright, contract, the Information Technology Act, 2000, confidentiality principles and, where personal data is involved, the Digital Personal Data Protection Act, 2023. With no statutory text- and-data mining (TDM) exception, AI training through large-scale scraping remains legally uncertain.
Policy and judicial trends increasingly favour licensed access, responsible data governance and transpar- ency over unrestricted extraction. 7.2 AI-Generated Works and Inventorship/ Authorship India currently adopts a human-centric approach to AI-generated works. Copyright generally subsists only where there is meaningful human creativity, with AI viewed as an assistive tool rather than an author. Purely autonomous AI outputs are unlikely to be pro- tected, whereas AI-assisted works involving substan- tial human input through prompting, editing or cura- tion may qualify, with ownership vesting in the human creator or employer, subject to contract. Similarly, AI cannot be named as an inventor under the Patents Act, 1970, and the IPO has rejected appli- cations identifying AI systems such as DABUS as inventors. Businesses therefore structure ownership through employment and assignment agreements, AI use policies, trade secret protection, and detailed records of prompts and human intervention to estab- lish the authorship, inventorship and ownership of AI- assisted creations. 7.3 Training Data, Model Development and Infringement Risk India has no dedicated statutory framework governing AI training data, leaving risks to be addressed through copyright, contract, trade secret and information tech- nology laws. The Delhi High Court’s ANI Media v Ope- nAI litigation is expected to determine whether the use of copyrighted material for AI training constitutes infringement. India does not recognise a TDM exception, creating uncertainty for developers using scraped copyright- ed content without a licence. Training on confidential information or trade secrets may also trigger breach of confidence, contractual and Information Technol- ogy Act claims. Enforcement currently centres on interim injunc- tions and dynamic and John Doe orders, with courts increasingly requiring the disclosure of dataset prov- enance and sourcing practices. Policy discussions favour a licensing-based framework balancing AI
125 CHAMBERS.COM
Powered by FlippingBook