The government’s reported plan to withdraw legal immunity from social media platforms could fundamentally change how companies such as Meta, Google and X operate in India. The Ministry of Electronics and Information Technology is examining whether platforms that use algorithms to rank, recommend and promote user-generated content should continue to enjoy safe-harbour protection under Section 79 of the Information Technology Act, 2000. The proposal, reported on November 4–5, is still at a preliminary stage and would require legislative changes.
There is a strong case for revisiting the existing framework. Social media companies exercise considerable control over what users see, even when they do not create the content themselves. Their recommendation systems can amplify misleading information, hate speech and harmful material while generating advertising revenue from user engagement.
READ | Legacy media survives social media onslaught amid information overload
But withdrawing safe harbour altogether would create problems of its own. India needs to make technology companies accountable for their role in distributing content without turning them into private censorship authorities.
Why social media algorithms need greater scrutiny
Social media platforms collect user data, predict viewing preferences and curate feeds to maximise engagement. Their algorithms influence which political opinions gain visibility, which misleading claims spread and which videos reach children. Although companies may not create the original content, their decisions about its distribution can substantially increase its reach.
The business model depends heavily on capturing attention. Content that provokes anger, fear or controversy can generate considerable engagement. Recommendation systems designed to maximise viewing time may repeatedly expose users to sensational or objectionable material. Platforms therefore cannot disclaim all responsibility for the commercial decisions that determine how content circulates.
The problem is particularly serious for children, who may encounter sexually explicit material, violent content or attempts at online grooming. Deepfakes and non-consensual intimate images pose another challenge. Once uploaded, such material can spread rapidly through recommendation systems and sharing features. Removing it after a complaint does little to reverse the damage already caused.
India has taken steps to address these risks. The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021 establish due-diligence and grievance-redressal obligations for intermediaries. Amendments notified on February 10, 2026, and effective from February 20, introduced specific requirements for synthetically generated content and shortened several compliance deadlines.
Under the amended rules, intermediaries generally have three hours to remove unlawful content following a valid court order or government intimation. The deadline for acting on complaints involving certain intimate imagery and impersonation has been reduced to two hours. The amendments also require labelling and other safeguards for specified categories of synthetic content.
Faster removal is necessary, but takedown rules alone cannot prevent harm. A misleading video may reach lakhs of viewers before it is removed. An intimate image may already have been downloaded and circulated elsewhere. Regulation must therefore address how platforms distribute and amplify content, alongside their obligations to remove unlawful material.
READ | How social media turns attention and data into profit
Why blanket liability threatens free expression
Making platforms legally responsible for everything their algorithms recommend could encourage excessive censorship. Faced with the possibility of litigation, companies may remove controversial material rather than examine whether it actually violates the law. Political commentary, satire, investigative journalism and legitimate criticism could become casualties of such caution.
This risk is particularly relevant in India, where political and social debates are frequently contentious. Article 19(1)(a) of the Constitution guarantees freedom of speech and expression, subject to reasonable restrictions under Article 19(2). Platforms should not be encouraged to suppress lawful expression merely because it attracts complaints or offends influential groups.
The Supreme Court’s 2015 judgment in Shreya Singhal v Union of India provides an important safeguard. While upholding Section 79, the court read down the provision concerning an intermediary’s knowledge of unlawful content. It held that the obligation to remove such material must arise from a court order or notification by the appropriate government or its agency, and must relate to the grounds specified in Article 19(2).
The court recognised the practical difficulty of requiring intermediaries to determine the legality of enormous volumes of content on the basis of individual complaints. That reasoning remains relevant as the government considers expanding platform liability.
There is also a risk to competition. Large technology companies possess the financial resources, legal teams and technical infrastructure needed to manage complex compliance requirements. Smaller platforms and start-ups have fewer resources. Treating every service that uses content-ranking technology as a publisher could raise entry barriers and strengthen the position of established companies.
Almost every major digital service uses some form of algorithmic ranking. Search engines order results, e-commerce platforms recommend products and video services suggest what users should watch next. The mere use of an algorithm cannot establish liability for everything it recommends. The relevant questions are how much control a platform exercises over distribution, what risks it can reasonably foresee and whether it takes proportionate steps to prevent harm.
READ | Karnataka’s social media ban idea misses the crisis in childhood
Conditional safe harbour offers a better approach
The European Union’s Digital Services Act provides a useful reference. It retains intermediary liability protections while imposing additional obligations, particularly on very large online platforms and search engines. These include assessments of systemic risks, measures to mitigate threats to children and electoral processes, and greater transparency over recommendation systems.
The Act also allows users of very large platforms to choose recommendation options that are not based on profiling. Companies covered by the law can face fines of up to 6% of their worldwide annual turnover for serious non-compliance.
India could adopt a similar risk-based approach rather than choosing between blanket immunity and blanket liability. Safe-harbour protection should remain available to platforms that comply with clearly defined duties appropriate to their role, scale and degree of control over content distribution.
Services that actively recommend, amplify and monetise public content should face greater obligations than those primarily transmitting private communications. Large platforms could be required to disclose the broad principles governing recommendation systems, assess risks to vulnerable users and demonstrate that they have reasonable safeguards against the amplification of unlawful material.
These obligations should be independently reviewable, with clear procedures for challenging wrongful takedowns. Smaller services should face proportionate requirements so that compliance does not become an advantage reserved for dominant companies.
The government’s reported consideration of stronger mechanisms for victims to seek removal of deepfakes and non-consensual images is welcome. Individuals whose images or identities are misused should have accessible remedies and prompt responses. But such mechanisms must distinguish unlawful abuse from lawful satire, criticism and other protected expression.
Section 79 was intended to support an open internet while requiring intermediaries to meet certain conditions. Its reform should recognise the responsibilities that platforms have acquired through algorithmic distribution. India would gain little by replacing a conditional legal protection with a system that makes companies fearful of carrying contentious speech. The better test is whether platforms can demonstrate reasonable care in the distribution systems they control.