Instagram tightens India feed: Meta says AI now keeps sexual and violent posts out of Rrecommendations
New Delhi: Meta said on Tuesday that it has adjusted how Instagram recommends content to users in India, using more advanced artificial intelligence models to keep sexually suggestive and violent posts from showing up in people’s feeds in the first place.
The company clarified that this is not a blanket ban. The changes apply only to content Instagram suggests from accounts a user does not follow, across Reels, Explore and Feed. Such posts are not being taken down simply for being recommendation-ineligible.
Fewer complaints, says Meta
In a post on its website, Meta claimed the tweaks have cut the rate at which Indian users report recommended content to about half of what it was a year ago. It did not share the actual number of posts affected, how much of the recommended content is now being filtered, or how it measured the drop in reports.
According to the company, the newer AI models check content directly against Instagram’s Recommendations Guidelines. This lets the system spot and limit questionable material on its own, instead of waiting for users to flag or hide it. “Today, our systems keep more of this content out of recommendations,” Meta said, naming sexually suggestive and violent content specifically.
Timing and the government advisory
The announcement comes as Indian regulators look harder at how social media algorithms work. On October 10, the Ministry of Electronics and Information Technology (MeitY) issued an advisory asking Meta to review its ranking and recommendation systems so that misleading, manipulated or out-of-context content is not amplified disproportionately. Meta did not say whether the new changes are connected to that advisory.
Recommendations vs. Community Standards
Meta stressed that its recommendation rules are separate from its Community Standards. Content that breaks the Community Standards is removed. Content that is borderline, sensitive or low quality, but does not break those rules, can stay on the platform; it just won’t be pushed to a wider audience. Material that is illegal in a given country is restricted there, the company said.
Controls users already have
Instagram also pointed to tools that let people shape their own experience:
Sensitive Content Control: lets users see less sensitive material across Search, Reels, Explore, suggested accounts and in-Feed recommendations. Meta said the most mature content is already excluded from recommendations in India.
Political content: users can opt to see less of it from accounts they do not follow.
Not Interested: flags a post so similar content is shown less often.
Hidden Words: filters chosen words, phrases and emojis from comments and message requests.
Block: stops an account from finding the user’s profile, viewing posts or sending messages, including through new accounts.
Protection for teens
Teen Accounts, launched in India last year, automatically place younger users in a 13+ content setting under Instagram’s age-appropriate guidelines.
Meta said signals such as hiding a post or tapping “Not Interested” also feed into its systems and added that it will keep refining its detection as user behaviour and technology change, guided by feedback from Indian users.
Also Read | Panchkula school assault: Doctors diagnose 8-year-old accused with conduct disorder and ADHD
- PTC NEWS