Magazine
Instagram slows sensitive Reels in India: complaints about recommendations halved
Meta is rolling out stricter AI filters for Instagram recommendations in India. According to the company, complaints about suggested content fell by roughly half within a year.

Contents · 3 sections
Instagram has tightened its recommendation systems in India – and reports roughly half as many complaints about suggested content as a year ago. In a statement dated 16 November 2026, Meta explains that improved AI models already review sensitive posts before they appear in the Reels feed, in Explore or as a recommendation in the main feed. Sexually suggestive content and depictions of violence are particularly affected.
For social media teams, this is an important signal: the change happens in the recommendation system. It therefore mainly affects organic discovery for people who don't follow an account yet.
AI decides earlier what ends up in recommendations
Instagram already had rules for which content is recommended to non-followers. The Recommendations Guidelines set an additional quality and sensitivity threshold for Explore, Reels and suggested posts. The Community Guidelines, by contrast, determine which content may stay online at all.
Meta now describes a changed order in moderation: more capable AI models assess posts directly, rather than waiting for reports or hidden posts. In India, these systems were specifically tuned to content that many users there have described as unwanted.
According to Meta, the result is that users in India report recommended content roughly 50 percent less often than a year ago. The company does not publish absolute case numbers.
For creators, the key distinction is between availability and recommendation. A post can remain visible on a profile and still be distributed less often beyond its own followers. On a platform whose Reels recommendations often prioritise very fresh videos, a restriction in the discovery surface carries particular weight.
India puts pressure on Instagram's ranking
The tightening also has a regulatory background. According to Reuters reporting dated 17 November, India's Ministry of Technology had asked Meta on 10 October to review its ranking and recommendation systems for amplifying manipulated, misleading or out-of-context posts. This also concerned content that could encourage violence or public unrest.
Meta presents the new filtering technology as the result of its own product development and user feedback. Even so, the sequence shows how closely algorithms, regional regulation and recommendation reach are now linked.
India is also a market where Meta is already testing new surfaces for Instagram: Srocket reported on the test of a Reels-centred start feed. The new moderation work targets exactly those surfaces where users discover content without having selected the sender beforehand.
What this means for brands
Meta does not announce a direct change to Reels distribution outside India with this India update. However, the company explicitly states that it is using more capable AI checks worldwide against unwanted recommendations, and that the additional fine-tuning took place in India.
For brands running international campaigns, the more important question is which motifs, topics and visual styles are suitable for recommendations in which market. Especially with controversial hooks, graphic images or provocative visual openings, reach among non-followers can differ from the reaction of the existing community.
Marketing teams should therefore not only compare likes and total views, but also look at the share of new viewers, reach outside the follower base, and the difference between paid and organic distribution. For campaigns aimed at young people, Instagram's globally expanded AI age detection adds an additional layer.
Users also have the familiar options "Not interested", Sensitive Content Control and settings for political recommendations. Meta treats these signals as feedback for personalised feeds. That Instagram is tuning its discovery surfaces more closely to quality and preferences fits with the development of more direct algorithm control on Facebook.
Sources: Meta – Instagram recommendations in India, 16. November 2026 · Reuters – context on the Indian government's request · Meta: recommendation controls and guidelines
Image: stock photo by Jakob Owens / Unsplash; not an illustration of the updated AI system.
Source


