Supreme Court of India proceedings on regulating AI and deepfakes (2025–2026)

Public interest petitions asking courts to regulate artificial intelligence have reached the Supreme Court of India under Article 32, which lets a petitioner approach it directly for enforcement of fundamental rights. Between 2025 and 2026 the Court has dealt with at least four such proceedings. Three are writ petitions filed in person by the same advocate, Narendra Kumar Goswami, and one is a transfer petition filed by the Union government. None has produced a reasoned judgment on whether the Constitution requires the State to regulate deepfakes or AI, and none has ordered the government to enact a law.

The petitions arrive against a backdrop of regulatory activity by the executive. The Ministry of Electronics and Information Technology (MeitY) has issued advisories on deepfakes, formed a Committee on Deepfakes in November 2024 on the Delhi High Court's urging, and amended the IT Rules in 2026 to cover synthetically generated information.⁹ They also arrive alongside parallel High Court cases. The Delhi High Court has been hearing deepfake PILs since 2023 (see the linked entry), and the Gujarat High Court took up a deepfake PIL in 2026.

The first petition, W.P.(C) 300/2025, was filed by an advocate who is a member of the U.P. Bar Council and appeared in person.¹ It asked the Court to direct MeitY to frame IT Act rules requiring labelling and watermarking of AI content, 24-hour takedown of deepfakes and algorithmic audits, and to create a National AI Regulation Authority. It also sought a Deepfake Monitoring Cell and AI-content rules in the Model Code of Conduct for the Election Commission, a national AI-threats protocol and police training for the Home Ministry, a deepfake literacy mission, declarations that the failure to regulate deepfakes violates Articles 14, 19(1)(a) and 21, and an expert committee.¹ The petitioner pointed to numerous fake videos of Colonel Sofiya Qureshi, the Army officer who gave media briefings during Operation Sindoor, circulating online.²³

On 16 May 2025 a bench of Justice Surya Kant and Justice Nongmeikapam Kotiswar Singh declined to entertain it. The order records that a PIL and connected matters on the same issues were pending before a Division Bench of the Delhi High Court headed by its Chief Justice, and that the Union had already responded there. It observed that one High Court was already seized of the subject, disposed of the petition, and gave the petitioner liberty to seek impleadment and assist in W.P.(C) 15596/2023 (Chaitanya Rohilla v. Union of India) and connected cases. It asked the High Court to hear him and consider his suggestions.¹

MediaNama reported the bench as calling the issue serious but cautioning against parallel proceedings, and quoted Justice Kant as saying that by the time one leaves the court "there will be another video".³ The result was that the Supreme Court did not itself address the legal framework and left the question with the Delhi High Court and, through it, with MeitY's committee.

Both of the petitioner's later writs were listed before the Chief Justice, Justice Joymalya Bagchi and Justice V. Mohana and disposed of without notice or any opinion on the merits.

W.P.(C) 823/2026 (11 August 2026). The petition asked for an urgent, court-supervised process on online harms, including threats of violence, doxxing, exposure of children's details, intimate or deepfake material shared without consent and harmful impersonation, with safeguards for political speech, journalism and criticism. Because the petitioner had already sent a representation dated 22 June 2026 to the ministries concerned, the Court disposed of the petition and directed the Union respondents and other stakeholders to examine the representation and take remedial measures as required by law. The order does not itself discuss AI or deepfakes.⁴

W.P.(C) 837/2026 (13 August 2026). This petition sought guidelines on the use of AI and similar tools by governments in welfare administration, surveillance, policing and online content moderation. It asked for declarations that deploying AI in governance without a statutory framework, human oversight and a right to explanation violates Articles 14, 19 and 21, a High-Powered Expert Committee to frame binding guidelines, an interim restraint on high-risk systems, impact assessments and bias audits of existing welfare, predictive-policing, facial-recognition and moderation systems, and an affidavit disclosing all such systems.⁶ The bench declined to entertain it as a writ and directed that it be treated as a representation. LiveLaw reported the Chief Justice as saying, "We are not the experts", and the bench as reasoning that the subject was highly technical and a matter of policy; it also noted that AI use in the judiciary is already regulated.⁶ The written order records that the respondents may consider the petitioner's suggestions and take appropriate measures, and that the petitioner may supplement his representation.⁵

The Supreme Court's only intervention so far in a pending High Court case on deepfake regulation is on the Union's transfer petition. Vikas Vijay Nair's PIL (R/W.P.(PIL) 9 of 2026) in the Gujarat High Court seeks a regulatory framework for AI-generated deepfake content and better coordination between police, platforms and the Centre's takedown portals. By an order of 10 April 2026 (reported on 15 April), a Division Bench of Chief Justice Sunita Agarwal and Justice D.N. Ray issued notice to Meta, Google, X, Reddit and Scribd, and as an interim direction asked intermediaries to onboard the Centre's SAHYOG portal and respond swiftly to Section 79(3)(b) notices. The Centre's affidavit said the main problem was inconsistent compliance by some intermediaries, not a lack of law or institutions.⁹¹⁰

The Union then filed Transfer Petition (Civil) No. 2597/2026 to move the case to the Supreme Court. On 8 September 2026, with Solicitor General Tushar Mehta appearing, the bench of the Chief Justice, Justice Bagchi and Justice Mohana issued notice returnable on 5 October 2026 and stayed further proceedings in the Gujarat PIL in the meantime.⁷ The short order records no reasons. According to LiveLaw, counsel for the Union said that similar proceedings before other High Courts had been interdicted and were pending before the Supreme Court; that SAHYOG, operational since October 2024, routes notices under Section 79(3)(b), with 524 intermediaries onboarded including Meta and Google; and that X had not onboarded and often failed to respond to notices.⁸

The stay froze a case in which X had applied on 14 August 2026 to modify the direction that it onboard SAHYOG, pointing to overlapping litigation, including an X appeal in the Karnataka High Court.¹⁰ The Gujarat case had been listed for 11 September 2026. Its stay therefore overlaps with the broader dispute over whether platforms can be compelled to accept executive takedown notices outside Section 69A, which the Supreme Court is also examining through the SAHYOG transfer petitions (see the linked entry).

Three themes appear in the orders, none of them yet a holding.

Policy domain and expertise. In August 2026 the bench treated the regulation of government AI and online harms as technical and a matter for the executive, not for a writ court that lacks the expertise.⁶ The effect is to leave regulatory design with MeitY and Parliament.

One forum. In 2025 the Court cited the principle against parallel proceedings and pointed the petitioner to the Delhi High Court.¹ In 2026 the Union asked the Court to centralise a similar dispute in itself, and the bench stayed the High Court case at the notice stage.⁷⁸ The Court has therefore used forum control, not merits rulings, as its main tool.

Relief through representation. Disposing of PILs as representations puts the petitioner's suggestions before the Union without a binding direction. The orders set no timeline for the government to respond.⁴⁵

These choices should be read against the Court's wider caution in technology cases. In Shreya Singhal v. Union of India it struck down Section 66A and read down Section 79, and in Puttaswamy it set the proportionality standard for State intrusions. No Supreme Court ruling located applies either decision to generative AI. The Union's position in the Gujarat matter, that the difficulty is non-compliance by intermediaries and not a gap in law or institutions, appears in that case's affidavits, and the Supreme Court has not tested it.⁹

The proceedings matter because they show who, in practice, will decide how India regulates AI and deepfakes. The Court has declined to take on the rule-writing role that the petitioner asked for, consistent with executive-led governance such as MeitY's 2026 amendments to the IT Rules. At the same time it has stayed a High Court case at the Union's request, which concentrates the forum for any later judicial scrutiny of takedown mechanisms in the Supreme Court.

The approach has costs and defenders. Disposing of a comprehensive PIL as a representation leaves the petitioner without a binding remedy and no deadline for the government, even as the bench in 2025 was reported by MediaNama to have acknowledged that those who produce such content move faster than regulatory mechanisms.³ Stays at the Union's request also pause High Court scrutiny of takedown notices that challengers say bypass the safeguards of Section 69A (see the linked SAHYOG entries). In favour of the Court's restraint, technical regulation is arguably better designed by expert bodies and Parliament than by a writ court, and the Union maintains that the difficulty lies in implementation, not in a gap in the law.⁹

For AI governance the central open question is whether the Supreme Court will hear a deepfake or AI-content case on the merits and, if so, how it will reconcile Article 19(1)(a) with intermediary duties under Section 79 and the 2026 IT Rules.

As of 15 October 2026, no Supreme Court judgment on regulating AI or deepfakes exists. The three Goswami petitions are disposed of. Transfer Petition (Civil) No. 2597/2026 was returnable on 5 October 2026; the order of that date, and whether the stay of the Gujarat High Court PIL continues, could not be located in the sources reviewed, and this entry does not state them.⁷ The Delhi High Court deepfake petitions were the main forum supervising the government's Committee on Deepfakes as of the last verified listing (see the linked entry).

Related Supreme Court proceedings on the SAHYOG portal (T.P.(C) 2071 to 2074/2026, stayed on 22 July 2026) and on police facial recognition at protests (the A.A. Rahim petition, taken up on 13 August 2026) are covered in separate entries.¹¹¹²