Content integrity

Instagram’s Feed first gathers posts from followed accounts and removes those that violate Community Guidelines, according to Meta’s Transparency Center’s June 2026 explanation1. During a later stage of selecting posts, it applies additional checks to reduce how widely problematic content is shown1. Recommendations have their own standards because they introduce material from accounts people have not chosen to follow: Meta’s Transparency Center says the Recommendations Guidelines help avoid suggesting low-quality, objectionable, sensitive or age-inappropriate content2. Instagram’s enforcement work also draws on shared company resources, which Mosseri explained in December 2020 included safety and integrity engineering teams across the company6.

Meta patents propose several ways to assess problematic posts and linked websites, rather than establishing which methods Instagram uses today78910. One Meta patent proposes comparing signs of objectionable material, including misleading or false information, against preset cutoffs to determine how prominently a post appears in a feed7. Another Meta patent proposes connecting factual articles with information about their authors, publishers and contents so a system can find similar or contradictory reporting and articles flagged as clickbait or editorialized9. For linked websites, a Meta patent proposes examining the web addresses a mobile device reaches after opening a link to detect sites that show acceptable content to the platform while concealing policy-violating material shown to visitors10. A further Meta patent proposes lowering the position of reshared links when fewer than a preset number of other websites link to their source, limiting their spread without blocking them8.

Instagram’s integrity decisions can also be examined through their effects on different communities, an issue Mosseri addressed in June 2020 when announcing a review of policies, tools and processes affecting Black people and other underrepresented groups11. He identified harassment, account verification, how widely content is shown and bias in automated decisions as the four areas of focus11. A separate Meta patent proposes evaluating content-policy changes before applying them broadly by comparing interactions with feeds containing content flagged under a rule against feeds containing only approved content, then using the comparison to decide whether to adjust the rule; this is a proposed method, not confirmation of Instagram’s review process12.

Instagram’s current protections distinguish personal accounts of suicide or self-harm struggles from prohibited material: people may share their own experiences, but those posts are hidden from teens even when the author is someone they follow, according to Meta’s Newsroom’s February 2026 account5. That account also describes support for searches concerning mental health more broadly, which can lead people to resources and helplines even when the search is not clearly related to suicide or self-harm5. At the most urgent end of this safety work, Meta’s Newsroom says it continues to alert emergency services when it becomes aware of someone at imminent risk of physical harm5.

Instagram’s integrity work also includes responses to exceptional information risks: Mosseri described election interventions intended to give people accurate voting information and reduce post-election uncertainty17. For crises more broadly, Meta’s Newsroom describes a company-wide Crisis Policy Protocol that assesses risks of imminent harm on and off its platforms and guides targeted policy responses18. Meta’s Transparency Center explains that heightened-risk responses can be limited, proportionate and time-bound, informed by monitoring real-world events, policy violations and emerging forms of abuse19. Those responses can include additional visibility restrictions on certain content that does not necessarily violate Meta’s rules, although this company-wide explanation is not an Instagram-specific list of measures19.