India Weighs Ending Social Media Safe Harbour as Algorithms Face Scrutiny

Legal protection for Facebook, YouTube and X faces a rethink

India's Ministry of Electronics and Information Technology is considering whether Facebook, Instagram, YouTube and X should lose the protection that generally shields them from legal responsibility for content posted by their users. The discussions centre on Section 79 of the Information Technology Act, 2000, and whether companies that choose which posts to recommend should receive the same protection as services that simply carry information.

The proposal remains exploratory. Officials are reviewing representations and legal feedback, with stakeholder consultations planned. No final decision or timetable has been announced. A broad withdrawal of protection from social media platforms would likely require an amendment to the IT Act and approval by Parliament.

The government's emerging approach could preserve protection for internet providers, telecom services and messaging services that deliver communications between users without interfering with their content. Platforms that rank, promote or suppress posts through algorithms could face a different legal standard.

For millions of users, the consequences could reach beyond the companies' legal bills. Greater liability could change which videos appear in recommendation feeds, how quickly disputed posts disappear and how platforms handle complaints about deepfakes or intimate images shared without consent. Creators and news publishers could also see changes to the systems that bring their work to an audience.

What Section 79 protects today

The government's published Information Technology Act calls Section 79 an exemption from intermediary liability in certain cases. In simple terms, it separates the person who posts something from the service that hosts or carries it, while setting conditions for that service to receive protection.

Section 2(w) defines an intermediary broadly. Its examples include telecom providers, internet providers, web hosting services, search engines, online payment sites, auction sites and online marketplaces. The law therefore covers a much wider group than social media companies alone.

Section 79(2) describes qualifying functions, including providing access to a communications system through which other people's information is transmitted, temporarily stored or hosted. It also sets out conditions concerning whether an intermediary initiates a transmission, selects its receiver, or selects or modifies the information transmitted. Intermediaries must observe due diligence and government guidelines.

Section 79(3) removes the exemption where an intermediary conspires in, aids or induces an unlawful act. It also addresses failures to remove or disable access to unlawful material expeditiously after the requisite knowledge or government notification, while preserving evidence. Court decisions have shaped how those notice requirements operate.

Companies already have obligations beyond taking down posts. Section 67C requires intermediaries to preserve prescribed information for the duration and in the form specified by the government. Intentional or knowing breaches can carry imprisonment of up to three years and a fine.

Why recommendation systems are central

Government officials are examining whether the way social media companies distribute content amounts to editorial control. Facebook, Instagram, YouTube and X host material created by users, but their automated systems also decide which posts appear prominently and which receive little attention.

An algorithm is a set of instructions used by software. In a recommendation feed, it can assess viewing habits, interactions and features of a post to predict what a person may watch or respond to. A video can consequently reach many people who never followed its creator.

The government's argument is that ranking, placement, promotion and suppression give platforms an active role in deciding a post's fate. Officials are also examining how companies apply their own community guidelines, which may differ from the requirements of Indian law.

A change built around recommendation systems would make the legal treatment of distribution especially important. The government would need to define which activities trigger additional responsibility and how that responsibility applies to unlawful material. Platforms, meanwhile, could reassess both the content they host and the content they recommend to a wider audience.

Child safety puts pressure on the current model

The Supreme Court's October 15 order in Just Rights for Children Alliance vs S Harish has become an important part of the debate. Addressing the circulation of child sexual exploitation and abuse material, the court called for a separate framework of proportionate penalties, including monetary fines.

The court identified a problem with relying heavily on the loss of safe harbour: it can be too drastic a response to routine failures while providing an inadequate deterrent against continuing violations. Its proposed approach would reserve withdrawal of protection for serious breaches and use other penalties for lesser failures.

The ministry's broader discussions and the court's proposed penalty structure therefore concern different regulatory choices. One focuses on which services should qualify for protection; the other concerns how to punish failures without using the most severe consequence in every case.

The Supreme Court has also sought responses from the Centre and Meta on a public interest petition filed by Sreerupa Chowdhury about algorithmic addiction and online harms to children. The petition raises concerns about infinite scrolling, autoplay videos, push notifications, inadequate age checks, online grooming and child sexual abuse material.

Those allegations bring product design into the discussion. Infinite scrolling continuously supplies new posts, while autoplay starts another video without a fresh choice by the viewer. The petition asks the court to consider safeguards against harms associated with these features, alongside protections against illegal content.

Political disputes have sharpened the confrontation

Relations between the government and Meta deteriorated after Facebook briefly withheld a video uploaded by Prime Minister Narendra Modi during student protests in July. The IT Ministry summoned Meta's global policy executives and raised concerns about compliance with Indian law and deepfakes.

Instagram's role in mobilising support during those protests also contributed to government scrutiny of recommendation systems. Officials have expressed concern that opposition or anti-establishment narratives sometimes gain greater visibility than government communication.

In August, Nishikant Dubey, the BJP MP who chairs the Parliamentary Standing Committee on Communications and Information Technology, warned that Meta could lose safe harbour. He said the committee had recommended withdrawing protection from platforms that fail to comply with Indian laws.

The scrutiny continued in October, when the ministry asked Meta to diversify narratives appearing on Facebook and Instagram amid protests against Chief Election Commissioner Gyanesh Kumar. An October 8 advisory also asked platforms to prioritise government complaints concerning public order and label misleading content.

In November, the Software Freedom Law Centre and content creators alleged that Meta restricted material about the Gyanesh Kumar protests through keywords or hashtags at the government's request. Journalists' accounts and protest posts were also restricted, with notices often offering little detail about the authorities or legal orders involved. These developments make transparency about restrictions a practical concern for people publishing political speech.

X's court challenge shows the stakes of removal notices

X has separately challenged the Centre's content blocking and removal framework, including its use of the Sahyog portal to issue notices. In September 2025, the Karnataka High Court dismissed the company's challenge, and X appealed.

The dispute concerns how government instructions reach platforms and the authority used to require action. Safe harbour adds another layer: a platform's response to a valid removal notice can affect whether it retains protection from liability for the material concerned.

The main developments span several months:

• July: The IT Ministry summoned Meta executives after Facebook briefly withheld Modi's video during student protests. The ministry also directed WhatsApp to pause its planned username feature pending consultations.
• August: Parliamentary committee chair Nishikant Dubey warned that Meta could lose safe harbour.
• September 2025: The Karnataka High Court dismissed X's challenge to the government's removal mechanism. X appealed.
• October 8: A ministry advisory addressed public order complaints and the labelling of misleading content.
• October 15: The Supreme Court called for proportionate penalties in the child protection case.
• November: The Software Freedom Law Centre and creators alleged restrictions on content about the Gyanesh Kumar protests.

WhatsApp's username plan raises a separate issue

WhatsApp's proposed username feature would let people contact one another without revealing their phone numbers. In July, the IT Ministry directed the company to halt the rollout pending consultations, citing concerns about impersonation, phishing and financial fraud.

Phishing involves tricking someone into revealing sensitive information or sending money, often by pretending to be a trusted person or organisation. Usernames can reduce the need to disclose a phone number, while also changing how people recognise and verify the person contacting them.

The government is considering a formal requirement for social media and messaging services to notify authorities before introducing major product changes that could affect user safety. Such a mechanism would give officials an opportunity to examine risks before a feature reaches users.

Messaging services consequently occupy two positions in the discussions. Services that simply carry messages could retain safe harbour under the proposed approach, while still facing scrutiny of new features and their safety consequences.

Faster complaints for people targeted by deepfakes

The ministry is also considering stronger ways for individuals to request removal of deepfakes, sexually explicit material shared without consent and other images or videos that misuse their identity. A deepfake is synthetic or altered media that makes someone appear to say or do something they did not.

The proposed mechanism would allow affected people to approach platforms directly and seek faster action. Officials want complaints from the person depicted to carry greater weight, rather than leaving the response largely to a company's internal moderation decisions or requiring the person to pursue a legal process.

For someone whose intimate image is circulating without consent, speed matters because copies can spread while a complaint is being processed. The practical issue is whether a victim can identify the material, reach a responsible team and obtain a timely decision.

India's Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, notified on February 25, 2021, already expanded due diligence requirements. They include grievance handling, information preservation, assistance to government agencies and removal of explicit content. The current discussions seek to strengthen the response available to affected users within that wider framework.

India's courts have already examined active roles

Safir Anand, senior partner at law firm Anand and Anand, discusses the balance between protection and responsibility in the firm's intermediary liability analysis. His account traces the role of decisions including Shreya Singhal v. Union of India in 2015 and Myspace v. Super Cassettes in 2017 in shaping the treatment of actual knowledge and removal obligations.

Anand argues that regulation should preserve room for lawful expression while requiring intermediaries to meet their responsibilities. He writes:

In this context, there needs to be a balance between granting exemptions and imposing liabilities.

The firm's analysis also examines Christian Louboutin v. Nakul Bajaj in 2018, where the Delhi High Court considered the difference between an intermediary and an active agent in online sales. Relevant activities included promotion, payment collection, quality assurance and guarantees of authenticity.

Another decision discussed in the analysis, Amazon Seller Services v. Modicare, recognised that providing services beyond basic hosting did not automatically remove the protection available to an online marketplace. For the social media debate, these cases show why the precise activity and the wording of any new legal test will matter.

Other jurisdictions offer different regulatory choices

The United States Supreme Court considered a dispute involving YouTube recommendations in 2023. The family of a woman killed in the 2015 Paris terrorist attacks argued that recommendations helped Islamic State spread propaganda and recruit supporters, and challenged Google's protection under Section 230 of the Communications Decency Act.

The court returned the case for reconsideration following its decision in a related case concerning assistance to terrorism. It left unresolved the broader question of whether algorithmic recommendations fall outside Section 230 protection.

The European Union's Digital Services Act takes an approach that includes duties directed at recommendation systems. It requires transparency about how recommendations work and user choices, with a recommendation option that does not rely on profiling for very large platforms. It also requires risk assessments.

A 2021 Court of Justice of the European Union ruling found that indexing, search functions and video recommendations alone were insufficient to establish an active role giving a platform knowledge or control of illegal content. India's deliberations therefore sit within a wider debate over how to assign responsibility for automated distribution.

For Indian users, the central choices are concrete: whether harmful material can be removed faster, whether platforms explain restrictions, and whether lawful speech loses visibility because companies fear legal exposure. The scope of any amendment, the proposed penalties and the complaint process will determine how those competing concerns are addressed.

Key Points

• India is exploring restrictions on Section 79 protection for social media platforms that use algorithms to determine content visibility.
• The discussions are preliminary. Broad changes would likely require Parliament to amend the IT Act.
• Services that simply transmit information, including qualifying messaging services, could retain protection.
• The Supreme Court has called for proportionate penalties, including fines, with safe harbour withdrawal reserved for serious violations.
• Separate proposals concern advance notice of major product changes and faster complaints over deepfakes and intimate images shared without consent.
• Political content disputes and X's challenge to removal mechanisms have intensified the debate over accountability and online speech.