Instagram removes hate speech aimed at a protected trait, not every insult. How to tell the difference and report so the review matches the right policy.
Instagram removes hate speech that directly attacks a person or group over a protected trait, such as race, religion, disability or gender identity, including slurs and calls for exclusion. A personal insult with no protected trait is usually bullying instead. Report it from the post, pick the matching category, and request a review if the decision seems wrong.
SummaryA comment calling you "stupid" and a comment calling your religion "stupid" look similar on screen, but Instagram reviews them under two different policies, and only one of them is what Meta classes as hate speech. Reporting under the wrong category is one of the most common reasons a report that should succeed does not.
Instagram removes hate speech that attacks a person or group over a protected trait, such as race, religion, caste, disability, sexual orientation, gender identity or serious disease. It generally leaves up a personal insult, a negative opinion, or criticism of a belief, policy or institution that does not target people on that basis. Knowing which side of that line a post sits on changes which report category gets it reviewed correctly.
This guide works through what Meta's hate speech policy actually covers, what it does not, how it differs from the separate bullying and harassment policy, and how to report so your complaint reaches the right reviewers the first time.
Last checked: 9 November 2026.
Instagram removes content that directly attacks a specific person or group based on a protected characteristic, meaning race, ethnicity, national origin, caste, religious affiliation, sexual orientation, sex, gender identity, disability or serious disease, through slurs, dehumanising comparisons, serious insults tied to that trait, or calls for exclusion. It generally allows criticism of ideas, institutions or policies, and personal insults that do not reference a protected trait, which fall under the separate bullying and harassment policy instead. Report the specific post or comment, pick the category closest to what actually happened, and if Instagram's first decision seems wrong, you can usually request a second review from inside the app.
Meta's Hateful Conduct policy defines hate speech as a direct attack on people, rather than concepts or institutions, on the basis of what it calls protected characteristics: race, ethnicity, national origin, disability, religious affiliation, caste, sexual orientation, sex, gender identity and serious disease. Age counts only when it is paired with one of those other traits, and the policy also protects refugees, migrants and asylum seekers from its most severe tier.
The policy's top tier removes several specific kinds of content when aimed at a protected trait: dehumanising comparisons to animals, insects or pathogens; allegations of serious criminality, such as calling a group predators or terrorists; calls or hopes for harm, including wishing disease or death on a group; harmful historical stereotypes; mocking hate crime victims; and slurs, which Meta defines as words that "inherently create an atmosphere of exclusion and intimidation" because of their link to historical discrimination or violence.
A second, slightly lighter tier covers calls for exclusion or segregation, and insults about a group's character, intelligence or worth, when they target a protected trait. Meta's own policy notes some deliberate exceptions, such as discussion of access to spaces commonly limited by sex or gender, and political or religious debate about topics like transgender rights or immigration, as long as the content stays in the territory of debate rather than an attack on people because of who they are.
A separate Bullying and Harassment policy covers attacks that are personal rather than group based. It says Meta removes content "meant to degrade or shame" private individuals, unwanted and repeated contact, threats to release private details, and claims about someone's sexual activity, character or appearance, none of which need to reference a protected trait to be actionable.
The two policies are reviewed by different teams against different rules, which is why the category you pick matters. A comment that says you personally are worthless is bullying. A comment that says people of your ethnicity are worthless is hate speech. Both can be removed, but only if each is reported under the policy that actually covers it.
| What was said | Likely policy | Why |
|---|---|---|
| "You're an idiot" | Bullying and harassment | Personal insult, no protected trait referenced |
| "People from [ethnic group] are all thieves" | Hate speech | Criminal allegation targeting a protected trait |
| "I disagree with this religion's teachings" | Usually neither | Criticism of a belief, not an attack on people |
| A slur aimed at your sexual orientation | Hate speech | Slur tied directly to a protected trait |
| A fake account made to mock one specific person | Bullying and harassment (and possibly impersonation) | Targets an individual, not a group |
| "People with [disability] shouldn't be allowed to work here" | Hate speech | Call for exclusion based on a protected trait |
Instagram's report a post or profile page lists "Violence, hate or exploitation" and "Bullying or unwanted contact" as separate categories for exactly this reason. Its harassment reporting guidance also covers an account set up specifically to target one person, which sits closer to bullying even if the content borrows hateful language.
A few patterns explain most rejections that are not simple policy disagreements. Meta's enforcement guidance notes that context matters: the same word can be a slur in one context and self-referential or condemning in another, and reviewers look at intent, not just the word used. Content that criticises a religion, a political position or an institution, rather than the people who hold it, is usually allowed even when it is harsh. Satire that clearly mocks the attacker's own view, rather than the targeted group, is also treated differently.
If a report is rejected and you believe the decision is wrong, Instagram's guidance on what to do if content you reported was not removed explains how to request a second review from Support Requests, usually reviewed again within about a day. Not every content type qualifies for a further appeal to the independent Oversight Board, and the board selects a limited number of cases.
A single hateful comment and an account that exists mainly to post that kind of content are different problems. If several posts from the same account show a pattern, report the profile itself, not only the latest post, and mention that it is a pattern if the form allows a note. Our guide to reporting a hate page targeting you specifically covers the personal-attack version of this, where an account is built around one individual rather than a group.
Platform rules and Malaysian law are separate questions, and a post can fail Instagram's rules without necessarily being a police matter, or the other way round. Where a post stirs up hostility along racial or religious lines at scale, it can fall within the remit of Malaysian authorities as well as Instagram's own policy. If the platform leaves a clearly rule-breaking post up after a proper report, MCMC's complaint channel is the route to escalate it, with your report reference attached. Direct threats of violence are a police matter regardless of whether the post also counts as hate speech.
Most individual comments are resolved by picking the right category and reporting once. A case review is worth considering when a coordinated group of accounts is targeting you or your business along the same lines, when reports keep being rejected despite matching the policy, or when the harassment overlaps with impersonation or a compromised account. SocialSafe by AwareXone reviews the case first, helps match the content to the policy it actually breaks, and prepares the report through Instagram's own categories. We never ask for a password or login code, and the platform makes the final call. Read how we work for the rest.
More free guides on the same kind of problem.
