Election Commission of India: Guidelines on Responsible and Ethical Use of Social Media Platforms (No. 491/SM_SOP/2024/Communication, 6 May 2024)

India's 2024 general election to the 18th Lok Sabha ran in seven phases from 19 April to 1 June 2024, with the Model Code of Conduct (MCC) in force from the announcement of the schedule on 16 March 2024. It was the first national election in which generative AI tools were widely available to campaigns and their supporters. Altered clips circulated widely during the campaign, among them videos of actors Aamir Khan and Ranveer Singh manipulated to appear to criticise the ruling Bharatiya Janata Party (BJP). Both actors filed police complaints. An edited video of Union Home Minister Amit Shah, falsely showing him promising to end reservations for Scheduled Castes, Scheduled Tribes and Other Backward Classes, led to police action against opposition functionaries in Delhi and Assam.⁵

At the end of April 2024 a BJP delegation led by IT Minister Ashwini Vaishnaw complained to the ECI about deepfakes. It asked that platforms be required to detect and remove election deepfakes faster and to block repeat offenders. MediaNama noted that the Amit Shah clip appeared to be conventionally edited rather than AI-generated.⁵ Earlier, in April, civil society organisations including Article 21 Trust, the Association for Democratic Reforms, the Internet Freedom Foundation (IFF), LibTech India and Maadhyam had written to the ECI. They pointed out that the platforms' 2019 Voluntary Code of Ethics had no legal force and asked the Commission to build an MCC for candidates and digital platforms through a transparent, participatory process, with clear enforcement and reporting.⁶

The immediate legal trigger was a public interest petition filed in the Delhi High Court by the organisation Lawyers Voice (W.P.(C) 6186/2024) against the Union of India through the Ministry of Electronics and Information Technology (MeitY), with the ECI as a respondent. Senior Advocate Jayant Mehta, for the petitioner, argued that the ECI already pre-certifies political advertisements in print and electronic media and should set up a similar mechanism for social media content.³ On 2 May 2024 the Division Bench of the Acting Chief Justice and Justice Manmeet Pritam Singh Arora declined to issue directions in the middle of an election. It recorded the ECI's statement that a comprehensive representation from the petitioner would be decided in accordance with law by 6 May 2024, directed the Commission to do so, and disposed of the petition.² According to press reports, the bench said the court had to trust the ECI to act. It also suggested that the ECI consider dynamic-injunction-style measures so that reposts of deepfakes could also be disabled. The ECI told the court that the deepfakes of Amit Shah, Rahul Gandhi, Aamir Khan and Ranveer Singh named in the petition had already been taken down.³

The letter, numbered 491/SM_SOP/2024/Communication and signed by Joint Director Anuj Chandak, went to the presidents, general secretaries and chairpersons of all recognised national and state political parties. Its subject was the responsible and ethical use of social media and the strict avoidance of wrongful use by parties and their representatives during the MCC period of the general election and by-elections.¹ The Commission recalled that its guidelines of 25 October 2013 had already extended the MCC to content posted online, including on social media, by candidates and parties. It warned that manipulated, distorted and edited content could sway voter opinion, deepen social divisions and erode trust in the electoral process, and that forwarding and re-sharing features made its spread dangerously unrestrained.¹ Citing the Delhi High Court's directions and the risk of parties, their representatives and star campaigners using deepfakes and AI-generated distorted content, it directed parties to observe the following:¹

1. No misinformation or synthetic deception. Parties must not use social media to spread misinformation or information that is patently false, untrue or misleading, or that impersonates another person. This expressly includes information synthetically created, generated or modified so that it reasonably appears authentic while actually deceiving recipients.
2. No impersonation of other persons, including other political parties or their representatives.
3. No content derogatory to women or repugnant to their honour and dignity.
4. No content involving children in political campaigning, in line with the ECI's earlier advisory against using children in campaigns.
5. No portrayal of violence, harm or harassment of animals.
6. No deepfakes on party handles. Parties must not allow their handles to publish or circulate deepfake audio or video that violates existing rules and regulations.
7. Three-hour takedown. When such deepfakes come to a party's notice, it must take the post down immediately, and in any case within three hours, and must identify and warn the person responsible within the party.
8. Reporting. Parties must report unlawful information and fake accounts imitating their official handles to the platform at once.
9. Escalation. If such content or accounts remain after reporting, parties may approach the Grievance Appellate Committee (GAC) under Rule 3A of the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021.

In a closing paragraph, the Commission directed all parties and leaders to refrain from using in their campaigns any technological or AI-based tools that distort information or spread misinformation and so lower the standard of electioneering.¹ The letter placed the obligations on political parties, not on platforms or AI developers, and set no new sanction of its own. Enforcement relied on the MCC and on the existing criminal and electoral offences listed in the letter.

The ECI did not create new offences. Instead it drew together the existing law that could apply to election deepfakes:¹

The Indian Penal Code provisions cited were replaced weeks later when the Bharatiya Nyaya Sanhita, 2023 came into force on 1 July 2024. The letter's description of prohibited content follows the vocabulary of the IT Rules, 2021 ("patently false", "untrue", "misleading", "impersonates another person"), a point critics later raised.⁷

The ECI was also acting against individual posts during the same week. On 7 May 2024 it wrote to X's nodal officer directing removal of an animated video posted by the BJP's Karnataka unit. It invoked Section 79(3)(b) of the IT Act and Rule 3(1)(d) of the IT Rules after X had not acted on an earlier notice from Bengaluru police.⁸

The Internet Freedom Foundation wrote to the ECI on 13 May 2024 setting out several criticisms.⁷

• Vague terms and satire. Undefined terms such as "misinformation", "patently false" and "misleading" echoed the contested language of the IT Rules and invited subjective, selective enforcement. The directions did not say how satire and parody accounts that "impersonate" parties would be treated.
• The three-hour deadline. Such a short deadline could cause panicked decisions. Cheap edits and even authentic media were often mislabelled as deepfakes, and AI-detection tools were neither fully accurate nor free of bias. IFF asked the ECI to assess realistically whether parties could reliably identify AI-generated media.
• Reliance on the GAC. Routing disputes to the Grievance Appellate Committee, a body constituted by the Union executive, made the government the arbiter of permissible online speech when an independent body was needed.
• Timing and scope. The directions came after two of the seven phases had already been polled. They also said nothing about communal content, despite complaints about such posts, including the BJP Karnataka video.

Other commentators saw the directions as a first step that did not reach the main channels through which manipulated content travels. These include unofficial and proxy accounts, influencers and paid promoters, which the civil society letter of April 2024 had identified as a gap.⁶ In a MediaNama discussion after the election, fact-checkers from Logically Facts, Newschecker and Quint Fact Check said deepfakes had not destabilised the 2024 election in the way some had feared. AI had been used more for campaigning and memes than for decisive disinformation. They added that public awareness of deepfakes had produced a "liar's dividend", in which politicians dismissed genuine clips as fake.⁹

The 6 May 2024 letter was the ECI's first instrument aimed specifically at deepfakes and AI-generated content. Several of its features became the basis of Indian election rules on synthetic media.

• Duties placed on parties. Rather than regulating platforms or AI tools directly, which falls within MeitY's remit under the IT Act, the ECI used its control over party conduct under the MCC to make parties responsible for what appears on their official handles.
• A fixed takedown clock. The three-hour deadline was carried forward in the ECI's advisory of 24 October 2025. That advisory extended it to any synthetic, AI-altered or manipulated content that misrepresents a person's identity, appearance or voice without consent and is detected on official party handles.¹¹ In April 2026 the Commission said that misleading or unlawful AI-generated content had to be acted on within three hours of being brought to platforms' notice.¹²
• From prohibition to disclosure. The 2024 letter only prohibited deceptive content. Its successors added positive transparency duties: labelling AI-generated material as "AI-Generated", "Digitally Enhanced" or "Synthetic Content" (January 2025), and labels covering at least 10% of the display area together with disclosure of the creating entity (October 2025).¹⁰¹¹
• Courts defer to the ECI. The Delhi High Court's refusal to frame its own guidelines mid-election, and its decision to leave the matter to the ECI, showed the courts' general reluctance to interfere with an election in progress.²³

The letter has not been withdrawn and remains part of the ECI's framework on campaign deepfakes. The ECI's 16 January 2025 labelling advisory described the May 2024 guidelines as the starting point of its approach and linked to them.¹⁰ The 24 October 2025 advisory was issued expressly in continuation of the communications of 6 May 2024 and 16 January 2025, enclosed both, and stated that its directions apply to all general elections and by-elections until further orders.¹¹ For the 2026 assembly elections in Assam, Kerala, Tamil Nadu, Puducherry and West Bengal, the Commission reiterated the labelling and three-hour requirements. It reported on 19 April 2026 that over 11,000 posts or URLs had been acted on since the elections were announced on 15 March 2026.¹²

As of 14 October 2026, the obligations still rest on the MCC and the ECI's general superintendence of elections rather than on any statute or rule specific to elections. No court has ruled on the validity of the three-hour requirement. The broader regulation of synthetic content on platforms is handled separately by MeitY under the IT Rules, 2021.