Reporting antisemitic content

This page names platforms, which nearly every other guide here avoids on purpose. Platform detail rots within months, so it carries a six-month review. It describes no menu path and no button: those move faster than anything written about them. Where this page and your screen disagree, your screen is right. What each is like to post to is platform by platform.

You need nobody’s permission, and you do not have to be Jewish or the target.

The honest frame: most reports go nowhere

Most reports produce nothing you can see. The content stays up, the account carries on, and you are told nothing was found. Report anyway, for three reasons that are not “it gets taken down”.

It builds a record. Volume against an account or a hashtag is what moves platforms and regulators. No single report does it, and the record does not exist unless people make it.

It creates evidence with a timestamp. This can become an employment matter, a university disciplinary or a police report, and a dated one beats your memory.

Some things do come down. The ones that do are reliably the ones reported precisely, under the right policy, with the breach named. That part is in your hands.

Before you report anything, capture it

Reporting can make the content vanish, and it takes your evidence with it. Do this first, every time. It is the most important instruction here.

1. The post, full screen rather than a crop of the words.
2. The account handle and display name, and the profile page.
3. The date and time.
4. The web address, saved separately as plain text.
5. What sits underneath: replies, accounts agreeing, share count.

Capturing evidence properly covers formats and storage. The fifth is the one people skip and later want: one post is an incident, forty is a pattern.

Report under the right category

This is where most reports fail, and it decides whether yours is read or closed unread. Nobody at a platform is assessing whether something is antisemitic. A reviewer or a classifier matches what you sent against a written policy, and “this is antisemitic” is not one. These are: hate speech, meaning an attack on people for a protected characteristic; violence and incitement; harassment of a person; terrorist content; and coordinated or inauthentic behaviour. Religion and ethnicity are protected on every platform below. The work is deciding which of those the content breaches.

What you are looking at	The policy it breaches
Jews control the banks or the media	Hate speech. Meta names this claim in its own standard.
Holocaust denial	Hate speech on Meta and YouTube. On X, abuse and harassment, as violent event denial.
Praise for an attack, or the group behind it	Violent and hateful organisations, not hate speech.
Abuse of one named person	Harassment. Where you are the target, report it yourself: X sometimes needs to hear from them.
Twenty accounts, one sentence	Coordinated or inauthentic behaviour, about accounts not content.

Where two apply, pick the more specific.

What to write where there is a box

Short, specific, in the platform’s own words. Not an essay about antisemitism.

Weak: “This is disgusting antisemitism and I cannot believe you allow it.”

Strong: “This post targets Jewish people with the claim that they control financial and media institutions, which your hateful conduct policy lists as a harmful stereotype it removes. The account posted three similar items this week.”

The rule underneath both: name the group targeted, name the policy, name the pattern. Three sentences at most.

Coded content, which is the hard one

A great deal of it is coded: a number, a symbol, a set of brackets, a phrase reading as ordinary if you have never seen it used. It means something exact to its intended audience and nothing to a reviewer abroad.

Say what the code means, in the report, in plain words. This feels like explaining the obvious and it is not. The person reading may genuinely not know, and Meta’s own standard says it treats certain words as proxies for protected characteristics based on local nuance.

“The brackets around the name are a marker identifying somebody as Jewish, used here to single this person out for the abuse in the replies.”

Where a code is contested, describe how it is used in this post rather than what it always means: a report that overclaims gets dismissed. And keep the explanation inside the report, because explaining a code publicly teaches it.

Platform by platform

X. Names religious affiliation as protected, and covers slurs, tropes, dehumanisation and hateful imagery, with the swastika and altered yellow star among its own examples. Holocaust denial sits under abuse and harassment instead. Bad at: most enforcement it lists reduces visibility rather than removing, so a report that worked and one that failed look identical. Appeal your own account, rarely another’s.

Instagram and Facebook. One rulebook, covering both and Threads, and the most specific policy here: it names Holocaust denial and the claim that Jewish people control financial, political or media institutions as stereotypes it removes. Quoting it back works well here. Bad at: comments and group posts, where much of this lives. Its exceptions for condemnation and satire catch your own post quoting the material to object to it. Appeal once.

TikTok. Covers hate speech, hateful behaviour and hateful ideologies, plus a ban on hateful organisations that extends to quoting them without disowning them. Good at: removing the obvious, at volume. Bad at: audio, on-screen text and irony. It also describes content staying up but no longer recommended, which is invisible to you. Appeals are quick. Use one.

YouTube. Covers denial of documented violent events, and conspiracy theories about a group having an agenda of world domination, a whole genre in itself. Strikes run against a channel, so a pattern beats one video. Bad at: comments and long uploads. Where the material is at minute thirty-four of fifty, say so: a reviewer made to hunt will not find it.

Discord, and chat or gaming platforms generally. Prohibits hate speech, violent extremism and harassment, enforces at account and server level, and may refer matters to the police. Good at: acting against a whole server, which no feed matches. Bad at: everything private, which is most of it, and messages go within seconds. Capture the message link as well as the image.

When to take it somewhere other than the platform

Three kinds of body, none named here: coverage differs by country, and naming one dates the page and reads as endorsement.

A national body recording antisemitic incidents. Most countries with a Jewish community of any size have one, often the main representative organisation or a community security body. Recording feeds the national figures, which is how the scale becomes arguable in public. Find yours through a synagogue or a Jewish student society.

A regulator. Some countries have an online safety or communications regulator with formal power over large platforms. What an individual can send one varies, and in many countries it is nothing.

The police. This page will not tell you what is criminal where you live. Thresholds differ by country and a confidently wrong summary is worse than none.

What each of those does with what you send, and what a reference number is actually for, are in after you report it.

Three questions for somebody local. Write down who answered and when.

1. Which body records antisemitic incidents here, and what happens to what I send?
2. What conduct here is criminal, and under which offence?
3. Can I report to the police online, and do they want the platform’s reference?

The routes go stale fastest. The offences move slowly.

Reporting as a group, and the trap in it

Do not organise a rota with a link in it. Coordinated mass reporting is visible and treated as manipulation. X’s authenticity policy prohibits coordinated inauthentic activity that artificially influences the platform, and a burst of near-identical reports from connected accounts is that pattern. Your own accounts can be actioned for it. It also hands the other side a screenshot of a group organising against one person, which reads as harassment whatever your intent.

What works is unglamorous. Several people who genuinely saw the thing, reporting independently, in their own words, at their own times, each under the category they judged right. No shared link, no template, no group message with a deadline.

Where somebody sends you a link asking you to report it, decide for yourself.

When it comes back “no violation”

The most common outcome, and not evidence you got it wrong.

Appeal once, where it is clear cut. Then stop. Log the outcome in one line and move on. Do not spend an evening on it, and do not open a second account.

Hold two possibilities. The review may have been perfunctory or automated. Or the category was wrong, and the same content reported as harassment rather than hate speech would have been actioned. Where three similar reports come back the same way, change the category.

Looking after yourself

An hour spent reporting antisemitic content is an hour spent reading antisemitic content, closely, and often about you. That is the real cost of this work and it appears in no outcome.

Cap the session before you start, at twenty or thirty minutes, with a timer somewhere you will hear it. Do it in daylight rather than at midnight. Once you have captured the replies, stop reading them.

Where this is aimed at you personally, when it comes at you is the page to be on first.

Make the folder tonight

Fill in the blanks145 words

Two questions from memory. What are the five items you capture before you open a reporting flow, and in what order? And what are the three things a free text box has to contain to make a moderator’s decision easy?

Then make the folder, and open a log with one line per report: date, platform, account, category used, outcome.

If I see something worth reporting, then I will capture the five items into the folder before I open any reporting flow.

If I am about to write in the free text box, then I will name the group targeted, the policy breached and the pattern, in three sentences.

And the obstacle, always the same one. If a report comes back with no violation, then I will log it, appeal once where it is clear cut, and close the tab rather than reading the account again.