Copyright in the digital environment and artificial intelligence

1. Software and databases

Computer programmes are literary works (section 2(o)), defined in section 2(ffc) as a set of instructions expressed in words, codes, schemes or any other form, including a machine readable medium, capable of causing a computer to perform a particular task. TRIPS Article 10(1) and WCT Article 4 require protection of programmes, whether in source or object code, as literary works. Protection extends to the code and to its literal and some non-literal elements, but not to the functional ideas, algorithms or methods of operation, which is the territory of patents (and section 3(k) of the Patents Act excludes a computer programme per se).

Databases are protected as compilations if they meet the Eastern Book Co. v. Modak standard of skill and judgment in selection or arrangement. India has no sui generis database right of the kind the EU created in its Database Directive (1996), which protects substantial investment in obtaining, verifying or presenting contents, regardless of originality.

2. Technological protection measures and rights management information

The 2012 amendment implemented WCT Articles 11 and 12 and WPPT Articles 18 and 19.

• Section 65A. Any person who circumvents an effective technological measure applied for protecting any of the rights conferred by the Act, with the intention of infringing such rights, is punishable with imprisonment up to two years and fine. Circumvention for purposes not expressly prohibited by the Act is allowed, and there is a list of permitted acts: encryption research, lawful investigation, security testing, protection of privacy, and measures for national security.
• Section 65B. Knowingly removing or altering rights management information without authority, or distributing copies knowing the information has been removed, is punishable similarly, and the owner may also bring civil proceedings.

3. Intermediaries and online infringement

Two statutes meet here: the Copyright Act (sections 51 and 52(1)(c)) and section 79 of the Information Technology Act, 2000, which gives intermediaries safe harbour if they observe due diligence and act on actual knowledge.

In Super Cassettes Industries v. MySpace Inc. (Delhi High Court Division Bench, 2016), the single judge had held MySpace liable for infringing content uploaded by users. The Division Bench reversed. It held that section 79 of the IT Act applies to copyright claims despite the proviso to section 81, that “knowledge” under section 51(a)(ii) means actual knowledge, not general awareness that some content might infringe, and that an intermediary must remove specific content on receiving notice identifying it. The court refused a general monitoring obligation, consistent with Shreya Singhal v. Union of India, 2015 INSC 257, on the meaning of actual knowledge.

Section 52(1)(c) separately allows transient or incidental storage for providing links or access, and requires a person providing it to take down the content for 21 days on a complaint from the owner, pending a court order.

Website blocking and dynamic injunctions

In UTV Software Communication v. 1337x.to (Delhi High Court, 2019), the court ordered internet service providers to block websites found to be rogue websites whose primary purpose is infringement, and allowed the plaintiff to extend the order to mirror and redirect sites through the Joint Registrar, without a fresh suit. The idea was borrowed from Singapore and UK practice. Such dynamic orders are now routine for films and sports events.

4. AI: who is the author of a machine output?

There are three positions.

1. Human authorship required. The US Copyright Office refuses registration to material generated by AI without sufficient human creative control, and in Thaler v. Perlmutter (D.C. Circuit, 2025) the court affirmed that the Copyright Act requires a human author. Prompts alone have been treated as instructions, not authorship.
2. The person making the arrangements. Section 9(3) of the UK Copyright, Designs and Patents Act, 1988 treats the author of a computer-generated work as the person by whom the arrangements necessary for its creation are undertaken. India’s section 2(d)(vi) says the person who causes the work to be created. These provisions were written for programmes run by users, not for systems that generate expression on their own, and it is unclear how the originality requirement applies.
3. No protection. Many scholars argue that outputs with no human creative choice should fall into the public domain, so that copyright continues to reward human creativity.

India’s Copyright Office registered an artwork listing an AI tool, RAGHAV, as co-author in 2020, and later issued a notice proposing to withdraw the registration. The episode shows the question is unresolved administratively as well as judicially.

5. AI: does training infringe?

Training a model involves copying works into a dataset. The questions are whether that copying is permitted by an exception, and whether a trained model or its outputs are infringing copies.

• EU. The Directive on Copyright in the Digital Single Market (2019) has text and data mining exceptions in Articles 3 (research organisations) and 4 (any person, unless the right holder has reserved the right in a machine-readable way). The AI Act (2024) requires providers of general-purpose models to have a policy for complying with that reservation and to publish a summary of training content.
• US. In 2025, two federal judges in the Northern District of California held that training on lawfully acquired books was fair use on the records before them (Bartz v. Anthropic and Kadrey v. Meta), while Bartz held that building a library from pirated copies was not. Other suits, including by news publishers, are continuing.
• UK. In Getty Images v. Stability AI (High Court, 2025), the court rejected Getty’s secondary infringement claim, holding that the trained model, which did not store the images, was not an infringing copy. The primary training claim had been dropped because training took place outside the UK.
• India. Section 52 has no text and data mining exception. In ANI Media v. OpenAI, pending before the Delhi High Court since 2024, the court is considering whether storage of news content for training infringes, whether outputs reproduce it, and whether an Indian court has jurisdiction over servers abroad. The Department for Promotion of Industry and Internal Trade has constituted an expert committee on AI and copyright.

How this is examined

Questions on “copyright and technology” or “challenges in the digital era” should cover software, TPMs and RMI, intermediary liability after MySpace, blocking orders, and the two AI questions. Mark the AI section as developing law and give the comparative positions.