New Delhi: Meta Platforms has strengthened Instagram’s content recommendation safeguards in India. It is deploying upgraded artificial intelligence models
to limit the visibility of violent and suggestive posts. The measures focus on content surfaced from accounts that users do not follow and apply to recommendations across Reels, Explore and Feed.
The company said the changes aim to prevent potentially sensitive material from reaching wider audiences. This step comes instead of removing all such posts from the platform. Meta announced the India-specific updates on Tuesday, but did not disclose the total volume of content filtered or the percentage of recommendations affected.
Instagram upgrades AI-based content filtering
The latest changes involve more advanced AI models that assess posts against Instagram’s Recommendations Guidelines. This lets the platform identify potentially inappropriate material and restrict its distribution before users flag it or hide it.
In a post on its website, Meta said the updated systems were keeping more suggestive and violent content out of recommendations. The company also claimed that the frequency of user reports about recommended posts in India has fallen by half compared with a year earlier.
Meta said it has been refining its default safeguards over the past 12 months, taking feedback from Indian users to identify the kinds of material they would prefer not to see. However, it has not explained the methodology behind the reported decline in complaints.
The update comes as Indian authorities increase their scrutiny of how social media platforms distribute content through automated recommendation systems. On October 10, the Ministry of Electronics and Information Technology issued an advisory asking Meta to examine its ranking and recommendation mechanisms. The government wanted platforms to ensure that potentially misleading, manipulated or out-of-context material was not being amplified disproportionately.
Meta has not confirmed whether its latest Instagram changes were introduced in response to the advisory. The company also maintains a distinction between its Recommendations Guidelines and Community Standards. While content that violates Community Standards can be removed from Instagram, material that is sensitive, objectionable or considered low-quality may remain on the platform if it does not breach those rules. Such posts can, however, be excluded from recommendations. Content that is illegal in a particular country is restricted in that market.
Apart from its automated filtering systems, Instagram provides several settings that allow users to manage the material appearing on their accounts. Its Sensitive Content Control feature lets users reduce the amount of sensitive material shown across Search, Reels, Explore, suggested accounts and Feed recommendations. Meta said its systems already prevent the most mature content from being recommended in India, while users can apply stricter settings to limit exposure further.
Other options include choosing to see fewer political recommendations from accounts users do not follow and selecting ‘Not Interested’ on posts they do not wish to encounter again. The Hidden Words feature can also filter specific words, phrases and emojis from comments and message requests.
Users can block individual accounts to prevent them from viewing their profiles and posts or contacting them. Instagram also offers additional safeguards through Teen Accounts, which were introduced in India last year. These accounts automatically apply a 13-plus content setting and age-appropriate guidelines for younger users.
Meta said feedback from users, including hiding posts and marking content as ‘Not Interested’, plays a role in improving its recommendation systems. The company plans to continue updating its detection technology as user preferences and online behaviour change.
















