Science and Technology › Computing and AI
हिन्दी — Read in HindiAI governance, safety and ethics
How artificial intelligence is governed: national and international rules, safety and the risks of frontier models, deepfakes, copyright, bias, liability and autonomous weapons. Prelims has asked about the AI Action Summit in Paris.
UPSC has asked
- Prelims 2025: the AI Action Summit held in Paris
All 7 subjects shown; the 2 that changed from 1 to 15 August 2026 are marked.Show only these
The European Union's AI Act
Copy link to The European Union's AI ActPrelims and Mains
The European Union's AI Act, in force since August 2024 and applicable from 2 August 2026, is the first comprehensive artificial intelligence law anywhere, and its method is risk based regulation: the duties on a system scale with the harm a use could do. Some uses are banned outright, such as social scoring; high risk uses, in hiring, credit, policing and infrastructure, must pass a conformity assessment before they reach the market; limited risk uses carry transparency duties, so content generated by artificial intelligence must be labelled or watermarked; the rest is free. General purpose models carry their own duties on documentation and copyright. It matters to India because Indian firms selling into Europe must comply, and because it is the template every other regulator argues with.
Unacceptable risk
- Some uses of artificial intelligence are banned outright
High risk
- A system needs a conformity assessment before it reaches the market
Limited risk
- Lighter checks apply
What changed
17 Aug 2026
- Anthropic said it will embed machine readable watermarks in Claude's text and files to meet the Act's transparency duty.
The Hindu, 17 Aug 2026: Anthropic watermarking of Claude AI content, explained (opens in a new tab)
14 Aug 2026
- An opinion piece argued that Europe's rules may become India's opportunity, as firms seek a jurisdiction with lighter compliance.
The Hindu, 14 Aug 2026: Europe's AI rules may become India's opportunity (opens in a new tab)
1 Aug 2026
- The Act became applicable on 2 August 2026, with an enforcement team launched the same day.
- Compliance deadlines for standalone high risk systems were pushed back to December 2027.
The Hindu, 1 Aug 2026: EU AI Act becomes applicable (opens in a new tab)
Deepfake labelling rules
Copy link to Deepfake labelling rulesPrelims and Mains
India reaches artificial intelligence through the platform that carries the content rather than the model that made it. The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021 place duties on intermediaries, with a significant social media intermediary defined as a platform with 50 lakh registered users or more, which must use automated tools against child sexual abuse material and the like. The amendment of 10 February 2026 brought synthetically generated information, audio, image or video created or altered so it appears real, under the rules: permitted AI content must carry clear labels and traceable metadata, takedown on a court order or a reasoned government intimation falls from 36 hours to three, a grievance must be settled in 36 hours rather than 72, and a sensitive case such as nudity or impersonation in two hours rather than 24.
What changed
6 Aug 2026
- The labelling duty and the three hour takedown were set out as the rules now stand.
PIB, 6 Aug 2026: Deepfakes: labelling rules and three-hour takedown (opens in a new tab)
30 Jul 2026Updatealso in news
- Thirteen responsible AI projects are running under the IndiaAI Mission, including Saakshya, a deepfake detection framework from IIT Jodhpur and IIT Madras.
Copyright in a machine generated work
Copy link to Copyright in a machine generated workPrelims and Mains
A copyright needs an author, and an author in Indian law is a person: under Section 2(d)(vi) of the Copyright Act, 1957 the author of a computer generated work is the person who causes it to be created. Fair dealing is the limited set of uses of a copyrighted work the Act allows without the owner's consent, such as research and reporting, and whether training a model on copyrighted text falls within it is the question the courts are now answering.
What changed
2 Sep 2026
- India's Copyright Office refused to name the AI system DABUS as an author, holding that an author must be a person.
24 Jul 2026Update
- In a case brought by a news agency against OpenAI, the Delhi High Court took the prima facie view that storing copyrighted work to train a language model falls within fair dealing.
- Its reason was that training supplies the work to no third party; it took the same view of outputs produced by retrieval augmented generation, and the ruling is interim.
Global governance of artificial intelligence
Copy link to Global governance of artificial intelligencePrelims and Mains
Governments argue less about which rules to write for artificial intelligence than about whether it needs rules written for it at all. Three positions have formed: the European route of a binding law scaled to risk; the United States position, in its Carolina Principles, that governments should avoid AI specific rules; and a bloc led by China offering developing countries an alternative framework built on cooperation and access rather than safety and ethics. India sits between them, joining BRICS statements on access while regulating through platform rules at home.
What changed
12 Sep 2026
- India joined the BRICS Leaders' Statement on the Global Governance of Artificial Intelligence, which sought stronger international cooperation and wider access to AI resources.
2 Sep 2026
- At the Group of Twenty the United States pressed its Carolina Principles, urging governments to avoid AI specific rules, as technology leaders warned against restrictions.
- Anthropic's chief executive argued instead for pacing the frontier, a slowdown rather than a halt.
20 Jul 2026Update
- China launched the World AI Cooperation Organisation with 29 other countries, including Russia and Pakistan, as an alternative for developing countries to Western frameworks built around safety and ethics.
A proposed law on AI
Copy link to A proposed law on AIPrelims and Mains
Artificial intelligence in India is governed through rules under the Information Technology Act, 2000, and the Ministry of Electronics and Information Technology has been weighing a separate law covering synthetic content, agentic systems and sandboxes for high risk uses.
What changed
22 Jul 2026New subject
- The Ministry of Electronics and Information Technology is weighing a separate law on artificial intelligence in place of rules under the Information Technology Act, 2000.
- Three strands are under study: consent for synthetically generated content, limits on how much autonomy agentic systems may have, and supervised sandboxes for high risk uses.
Rome Declaration
Copy link to Rome DeclarationPrelims and Mains
The Rome Declaration is a call by Nobel laureates, AI scientists and religious leaders for a treaty keeping autonomous systems out of nuclear launch decisions, on the principle of meaningful human control over the use of lethal force.
What changed
23 Jul 2026New subjectalso in news
- Nobel laureates, AI scientists and religious leaders called for a treaty barring autonomous systems from nuclear launch decisions.
Export controls on frontier models
Copy link to Export controls on frontier modelsMains
Export controls on artificial intelligence work on scarce hardware such as advanced chips; a model whose weights have been published cannot be withdrawn, which limits what controls on models themselves can do.
What changed
3 Jul 2026New subjectalso in news
- The United States classed the most advanced models of one company as dual use cyber tools and restricted foreign access, and a Chinese firm released a rival open weight model within a day.
- Controls work on scarce hardware, but published model weights cannot be withdrawn.
The Hindu, 3 Jul 2026: Can caged frontier AI models outrun open ones? (opens in a new tab)