Meta details changes to political content recommendations after scrutiny of algorithms in India

New Delhi

Meta has detailed changes to Instagram’s recommendation systems in India, bringing renewed attention to the way social media algorithms determine which political narratives reach users, how frequently they appear in individual feeds and how their visibility expands beyond their original audiences. The announcement, published on 16 November 2026, comes after the government sought a review of the company’s ranking and recommendation mechanisms, raising questions about the influence exercised by social media platforms over the distribution of political content.

Read had investigated this aspect of social media platforms in August this year, examining how Meta’s recommendation algorithms influence political mobilisation and the distribution of political narratives among Indian users. The investigation, titled “How Meta’s platforms are rewriting political mobilisation”, highlighted an important aspect of algorithmic amplification: while the actual distribution of content is automated, the systems do not operate independently of human decision-making.

You Might Be Interested In

The investigation established that Meta’s engineers design and update recommendation systems, determine the objectives they optimise for and establish the policies and parameters governing which content becomes eligible for recommendation. In other words, algorithms may decide which post appears on an individual’s feed, but the framework within which those decisions are made is designed and controlled by people within the company.

The investigation also examined how two individuals following similar political accounts could receive substantially different recommendations because Facebook and Instagram continuously personalise content according to predicted user behaviour. The systems assess several signals, including previous activity, viewing time, shares, comments, relationships and predicted relevance, before determining what is likely to appear in an individual’s feed.

Another important finding concerned the relationship between paid political advertising and organic algorithmic amplification. Political campaigns can purchase initial visibility through advertisements, following which engagement generated through shares, comments and viewing time can help content reach audiences beyond the original advertising target. This combination of paid promotion, user engagement and algorithmic recommendations has significantly expanded the ability of political campaigns to acquire digital visibility.

The government’s intervention on the issue came on 10 October 2026, when the Ministry of Electronics and Information Technology issued a specific advisory to Meta, asking it to review the content-ranking and recommendation systems used by Facebook and Instagram. The ministry asked the company to ensure that its recommendation systems did not result in disproportionate exposure of users to a single or limited set of narratives and that they provided an appropriate diversity of relevant and credible information.

The advisory also asked Meta to review its ranking mechanisms to prevent potentially misleading, manipulated or out-of-context material from being disproportionately amplified or repeatedly recommended to users. It called for stronger mechanisms to identify content acquiring significant reach through recommendation systems, particularly material that could create public confusion, apprehension or disturbance of public order.

The advisory was significant because it addressed not merely what content social media platforms allow users to upload, but how their algorithms determine what users are repeatedly shown. The distinction is important because material that remains available on a platform can acquire substantially greater visibility through recommendation systems, even when users have not actively sought it out.

In its announcement on 16 November 2026, titled “An Update on the Content We Recommend on Instagram in India”, Meta outlined improvements to its recommendation systems. The company said it had strengthened its default protections over the preceding year and improved its detection mechanisms, including through more capable artificial intelligence models that assess content before it is recommended rather than waiting for users to report or hide it.

Significantly, Meta also detailed its Political Content Control feature, which allows users to choose to see less political content in recommendations from accounts they do not follow. The feature is available under Settings and Activity, within Content Preferences, under Political Content. This provides users with a mechanism to adjust the amount of political content appearing in their recommendations.

Meta has said that improvements to its recommendation systems have been underway over the preceding year and have been guided by user feedback.

The announcement nevertheless comes at a time when the functioning of recommendation algorithms has become a specific subject of regulatory scrutiny in India. The government’s October advisory and Meta’s subsequent explanation of its recommendation systems have brought greater attention to how platforms determine the visibility and distribution of information.

In a subsequent investigation published on 8 November 2026, Read examined the government’s consideration of changes to the legal framework governing social media platforms, including the safe-harbour protection available under Section 79 of the Information Technology Act. The investigation examined whether companies that actively rank, recommend and amplify content should be treated differently from intermediaries that merely transmit information uploaded by users.

Leave a Comment