The government is considering withdrawing blanket protection against liability for user-generated content currently available to social media platforms under Section 79 of the Information Technology (IT) Act, 2000, Business Standard reported. 

The proposal would target platforms that use algorithms to determine the ranking, placement, promotion or suppression of content, while potentially exempting peer-to-peer messaging services that do not interfere with messages transmitted between users.

“Any social media intermediary that uses algorithms to decide the ranking, placement, promotion or suppression of user-generated content is no longer just a pipeline to ferry such content but has an active role in deciding its fate. A blanket safe-harbour protection against liability for such content, therefore, needs a relook,” a senior government official told the publication.

What is safe harbour under Indian law? Section 79 of the IT Act provides intermediaries with conditional immunity from liability for third-party information, data or communications hosted or transmitted through their services. To qualify, platforms must not initiate transmissions, select recipients or modify the information being transmitted, while also complying with prescribed due diligence obligations. The provision, however, does not protect intermediaries that abet unlawful acts or fail to remove unlawful content after receiving the requisite notice from the government or a court. 

Government discusses algorithmic liability: The discussions are at a preliminary stage and will require an amendment to the IT Act. The government is reviewing representations and legal feedback on the issue and plans to consult stakeholders before proposing any changes.

The proposal follows concerns about platforms actively influencing the content users see. Under the existing framework, intermediaries can lose safe harbour protection if they fail to comply with due diligence requirements. However, the proposed withdrawal would extend the scope of potential liability to platforms that algorithmically determine content visibility.

“If an algorithm decides what a user should be shown, the intermediary cannot escape its liability. Who is to be held responsible if explicit content is inadvertently shown to a social media user below the age of 18?” the official said.

Direct takedown mechanism for deepfakes: Separately, the government is considering a mechanism that would allow users to directly approach social media platforms to request the removal of deepfakes and non-consensual images or videos that directly impact or relate to them.

Currently, users generally have to approach law enforcement agencies or obtain court orders to secure the removal of such content. The proposed mechanism would allow affected individuals to seek takedowns directly from the platform. 

“Currently, the removal of most such content requires a legal process, under which users have to file a complaint with law enforcement agencies or obtain a court order. In cases involving deepfakes or non-consensual images and videos, time is of the essence. Users must therefore be given the option to seek their removal from platforms as quickly as possible,” the official told Business Standard. 

Why does this matter?  The proposed withdrawal of safe harbour protection for algorithmic recommendations could increase the pressure on social media platforms to suppress content critical of the government. 

In October, MeitY asked Meta to diversify narratives appearing on Instagram and Facebook amid protests against Chief Election Commissioner Gyanesh Kumar, while its October 8 advisory asked platforms to prioritise government complaints about content affecting public order and to label “misleading content”. 

Subsequently, Meta restricted journalists’ accounts and protest-related posts, with notices often providing little information about the legal orders or authorities involved. In November, the Software Freedom Law Centre and content creators alleged Meta was restricting content regarding the Gyanesh Kumar protests through keywords or hashtags at the government’s request. 

Removing safe harbour for algorithmic decisions could incentivise platforms to pre-emptively censor contentious user-generated content, including by shadow-banning posts, deprioritising critical narratives or limiting their reach through recommendations. Consequently, political speech, journalism and documentation of alleged police brutality could face restrictions without outright takedowns, leaving users with limited visibility and avenues for redress.

Also read: