
Meta says it has adjusted Instagram’s recommendation systems for India to reduce how often people encounter unwanted sexually suggestive and violent content. The change concerns what Instagram recommends, particularly from accounts a person does not follow. It is not an announcement that every such post has disappeared from the service.
For users, the important distinction is between content being available on Instagram and content being actively suggested to a wider audience. Meta’s update describes changes to that second layer, alongside controls people can use themselves.
Recommendations are different from removals
Instagram suggests posts and videos in places such as Explore, Reels and Feed. Those recommendations can introduce accounts and subjects beyond a person’s existing follows.
Meta says these suggestions are subject to its Recommendations Guidelines, which are intended to limit the promotion of content it considers low quality, objectionable or particularly sensitive. Separately, the company says content that violates its Community Standards is removed, while content unlawful in a country is restricted there.
That separation matters. A post can be unsuitable for recommendation without being treated as content that must be removed. Reducing its reach through suggestions is therefore not the same as imposing a platform-wide ban on it.
What Meta says it changed in India
The company describes two strands of work: improvements to detection globally, including more capable AI models that assess content before it is recommended, and changes tuned specifically to feedback from users in India.
According to Meta, its India-focused changes keep more sexually suggestive and violent material out of recommendations. The company presents this as work carried out over the past year, rather than a promise that unwanted content can no longer appear.
There is also a measurement caveat. Meta says people in India now report recommended content about half as often as a year earlier. Reuters reported the company did not supply the underlying figures.
Reporting frequency is not the same measure as the prevalence of harmful content. Without the underlying data and measurement details, that statement should not be rewritten as proof that harmful content itself has fallen by 50 percent. It is a company-reported change in how frequently recommendations are reported.
Which controls do which job?
Meta’s update points to several controls, but they address different problems:
- Sensitive Content Control: adjusts the amount of sensitive content in recommendations. Meta places it under Content Preferences in Settings and Activity.
- Political Content Control: lets people choose to see less political content recommended from accounts they do not follow. Meta lists it under Content Preferences, then Political Content.
- Not Interested: provides feedback on a recommended post or reel through its three-dot menu.
- Hidden Words: filters selected words, phrases and emojis in comments and message requests. This is not the same control as filtering the recommendation feed.
- Block: limits another account’s ability to find a profile, see its posts or send messages, and can cover new accounts that person tries to create.
These are Meta’s described controls, not a guarantee of a perfectly clean feed. A recommendation setting, a comment filter and blocking a person are separate tools and should not be treated as interchangeable.
What the update does not establish
Meta also points to additional age-appropriate protections for teens. Those sit alongside the broader recommendation changes; the India update is not limited to teen accounts.
Reuters placed the announcement in the context of increased government scrutiny of recommendation systems. But it also reported that Meta did not link the changes to the government advisory. Timing alone does not establish that an advisory caused a specific product change.
The practical takeaway is narrower: Instagram says it has changed which posts it recommends in India, while retaining separate rules for removal and separate tools for personal preferences. Users can check those controls, but the company has not demonstrated in this announcement that unwanted content has been eliminated.
This explainer separates Meta’s statements from independently established outcomes. It does not claim an independent audit of Instagram’s recommendation systems.