The BRICS New Delhi Declaration commits its members to “promoting respect for the intellectual property rights used in the digital environment, including for artificial intelligence training purposes, as well as fair remuneration to right holders, while respecting the needs and priorities of developing countries”. The Rio de Janeiro Declaration of July 2025 says the same thing, in the same words. India held the chair this year, and the sentence did not change.

Why it matters: A great deal happened in India in the intervening year. A single judge of the Delhi High Court decided ANI Media Ltd. v. OpenAI OpCo LLC on July 24, 2026, refusing an injunction against training on the news agency’s material. The appeal is live and the Delhi HC has sought a response from OpenAI. DPIIT’s (Department for Promotion of Industry and Internal Trade) Part II working paper on copyright in AI outputs, promised for February 2026, has not been published. The Copyright Office decided the DABUS application. None of it produced a new word in the paragraph the Indian chair took to its own summit.

What paragraph 120 does add

Two things are new compared with Rio. The first is a line on cooperation among BRICS intellectual property offices, which now extends to “AI application” alongside patent examination, traditional knowledge and geographical indications.

The second is a sentence that has no equivalent in the Rio text:

“With the rise of application of AI, we recognize risks relating to misappropriation and misrepresentation of knowledge, heritage, and cultural values that are insufficiently represented in data sets and AI models.”

That is the one place where the declaration says something about AI and creative work that it did not say last year. It is a recognition of a risk, not an obligation on anyone, and it names no mechanism for addressing it. But it is the sentence a country arguing for training-data transparency would build on.

Prof. Nikhil Naren, Assistant Professor at Jindal Global Law School and Assistant Director of the Cyril Shroff Centre for AI, Law and Regulation, said the concern is not new, and that Indian law offers fragmented protection while collective cultural knowledge often falls between its categories. The greater difficulty, he said, is that cultural harm can occur without conventional copying: a model might misrepresent a community’s traditions, strip knowledge of its context, or commercialise a recognisable aesthetic without infringing any identifiable copyright. He said he would support community participation in decisions about digitisation and training, records of where material came from, attribution, benefit-sharing where justified, and practical correction mechanisms.

“The declaration points towards remuneration, but cannot make a use protected by fair dealing payable”

Prof. Naren told MediaNama that paragraph 120 cannot do what its language gestures at. A declaration cannot make payable a use that Section 52 protects.

He also said the judicial position is still moving, and that it would be inaccurate to present either mandatory licensing or unrestricted training as settled. The Delhi High Court’s interim ruling of July 2026 accepted the fair-dealing argument for the training use before it, and that ruling is under appeal.

His larger concern is a problem the copyright test does not reach at all:

“An AI service may compete with the people whose work helped develop it without reproducing a recognisable passage from any particular work. That does not automatically establish copyright infringement, but it exposes a policy problem that an output-copying test alone cannot resolve.”

Any remuneration framework, he said, should ensure that independent authors and smaller publishers actually receive a fair share, with transparency about what was used and how payments are calculated.

The one AI paragraph that hardened

Paragraph 81 is where the modality changes. On the BRICS Leaders’ Statement on the Global Governance of Artificial Intelligence, adopted at Rio on July 6, 2025, the New Delhi text says: “We commit to implement”.

That is a real shift, and it is the only place in the declaration where AI attracts a commitment to implement anything. What it commits India to is a question worth asking, because the Rio statement says its guidelines are to be applied “through either domestic or applicable international frameworks”. Implementation, in other words, is domestic. Which returns the question to Ministry of Electronics and Information Technology (MeitY) and DPIIT.

Naren said a declaration of this kind can support an argument before a court but cannot override Section 52 or independently create a royalty obligation. Its sharper significance, on his reading, is as an accountability benchmark: it gives creators grounds to ask how India’s domestic policy delivers on an international commitment to remuneration.

Paragraph 81 also congratulates India on the AI Impact Summit held in February 2026 and acknowledges the World AI Conference and High-Level Meeting on Global AI Governance in Shanghai.

Where else AI appears

AI is named in 11 of the declaration’s 140 paragraphs. Beyond paragraphs 81 and 120:

  • Paragraph 59 — health. Leveraging digital health technology and AI in health infrastructure, surveillance, regulation, research and capacity-building.
  • Paragraph 65 — energy. AI and advanced data analytics in system planning, grid management and renewable integration, with BRICS Guiding Principles on Smart Grids and Energy Storage, and a proposed Digital Centre of Excellence.
  • Paragraph 72 — industry. The BRICS Centre for Industrial Competencies, set up with UNIDO, facilitating exchanges on frontier technologies including AI.
  • Paragraph 95 — finance. Sharing experience on cybersecurity, AI and quantum computing for the financial sector through the BRICS Rapid Information Security Channel and the BRICS Fintech Working Group.
  • Paragraph 100 — a Discussion Paper on Artificial Intelligence and the Green Economy, and technical notes on the macroeconomic implications of AI, under the new BRICS Task Force on Growth and Development.
  • Paragraph 118 — standards. A BRICS Technical Session on Emerging Trends in AI Standardization, and continued exchange among national standards bodies.
  • Paragraph 119 — statistics. AI among the innovative methodologies and data sources in national statistical reform.
  • Paragraph 125 — education. “Safe, ethical, human-centred use of emerging technologies, including Artificial Intelligence, in education.”
  • Paragraph 129 — tourism. AI and digital tools for “smarter, more efficient tourism systems”.

Two more paragraphs sit next to AI without naming it. Paragraph 42 links cross-border scam networks to digital technologies and payment systems and calls for cooperation on digital financial security. Paragraph 48 lists deepfakes, alongside misinformation and disinformation, among the misuses of technology that BRICS members will address.

What none of it contains

No definition of AI. No timeline. No reporting requirement. No named body accountable for any of it, except the working groups the declaration itself created. Across 11 paragraphs, the verbs are recognise, note, acknowledge, welcome and encourage. “Commit to implement” appears once.

For a beat that spends its time distinguishing what is in force from what is proposed, a summit declaration sits below even a draft rule. Its value is as evidence of a position. On AI and copyright, the position India took to its own summit in 2026 is the position Brazil took to its summit in 2025.

On the repetition of the 2025 language, Naren said it suggests agreement on an aspiration while leaving the distribution of benefits unresolved. Who receives remuneration, he said, matters as much as who pays it, and a system that compensates large rights holders without reaching individual creators could reproduce existing inequalities. He said he is increasingly sceptical of broad assurances that AI will benefit everyone while the distribution of those benefits remains unclear. India has both an interest in AI development and a substantial creative economy, and its position becomes clearer only when those institutional choices are made.

Questions MediaNama has sent

To DPIIT:

  • Whether the position in paragraph 120 reflects any settled Indian view on AI training and remuneration to rights holders.
  • When Part II of the working paper on copyright in AI outputs will be published.

To MeitY and the Ministry of External Affairs:

  • Which ministry is responsible for implementing the BRICS Leaders’ Statement on the Global Governance of Artificial Intelligence, and what implementation involves.
  • Whether India proposed the new sentence in paragraph 120, and whether any mechanism was agreed for the risks it identifies.

This copy will be updated with the responses.

Updated on 16 September 2026 at 10:46 am with comments from Prof. Nikhil Naren.

Also read