Account safety on X
X (Twitter) Under the Hood labels, explained: what each one does to your posts
Every label X’s Under the Hood report can show, in plain words: what it means, who still sees your posts, and what to do about it. From X’s own code.
Short answer: X’s Under the Hood report can show 19 labels on posts and 12 on accounts, plus a few for legal demands and copyright. Most of them do the same thing: your followers still see the posts, and X stops recommending them to anyone else. A few go further. Some put a post behind a warning, the FOSNR labels keep a post on your profile with a label everyone can see, and SPAM, PDNA and BOUNCE take a post off X. Drop your report into our free Under the Hood reader to see your own labels in plain words.
What Under the Hood is
Since August 13, 2026, X lets you see the labels it put on your posts and your account last month: the labels that can limit who sees them. You’ll find the report at x.com/i/under_the_hood on a computer, for accounts at least a year old with 10 or more posts in the month before. How to get it, and what “None in this period” means, is in X shadowban: how to see yours with Under the Hood.
This article covers what comes after: what each label in the report means. Every label below comes from the list in X’s own open-source code, read in October 2026.
How to read the numbers
SPAM_HIGH_RECALL3 of 40 posts · 7.50%
NSFW_TEXT1 of 40 posts · 2.50%
Account labels
SpamHighRecall5 of 31 days · 16.12%
Download this report (JSON)
Each label in the report comes with a count:
- On a post label, the count is posts: “3 of 40 posts · 7.50%” means X put that label on 3 of the 40 posts you made that month.
- On an account label, the count is days: “5 of 31 days · 16.12%” means the label sat on your whole account for 5 days of the month.
The report doesn’t say which posts got a label. X also gives two lines for each label, “What it means” and “Effect on visibility”. That second line is the one to read first.
Post hidden from recommendations to non-followers.
That’s the effect of SPAM_HIGH_RECALL, and of most of the other labels too.
What each label does to your posts
| What happens | Who still sees the posts | Labels |
|---|---|---|
| Hidden from strangers | Your followers | Most labels, such as SPAM_HIGH_RECALL and SpamHighRecall |
| Behind a warning | Your followers, after a warning; not people under 18, people with no age set, or anyone logged out | NSFW_HIGH_PRECISION, GORE_AND_VIOLENCE_HIGH_PRECISION and others |
| Only on your profile | Anyone who opens your profile, with a label on the post | FOSNR_ABUSE, FOSNR_HATEFUL_CONDUCT, FOSNR_VIOLENT_SPEECH, FOSNR_CIVIC_INTEGRITY |
| Not shown on X | Nobody | SPAM, PDNA, BOUNCE, Dmca |
| The account is held | Your followers, while you act | ReadOnly, Compromised, AbusiveHighRecall |
Spam labels
| Label | On | In plain words |
|---|---|---|
SPAM_HIGH_RECALL | Posts | X’s spam filter thought these posts might be spam. Followers still see them; X stops recommending them to anyone else. |
SPAM | Posts | X decided these posts broke its authenticity policy, which covers spam. They aren’t shown on X. |
SpamHighRecall | Account | X’s systems thought the account was likely to post spam. On those days, none of its posts were recommended to people who don’t follow it. |
These are the labels a reply habit can earn. X’s spam rules name bulk, aggressive replies, many replies to one person, and the same words again and again. We lost an account to that kind of pace, and tell the story in We lost a Twitter (X) account because of replies.
Adult and violent content labels
| Label | On | In plain words |
|---|---|---|
NSFW_HIGH_RECALL | Posts | X thought these posts might contain adult content. Hidden from strangers, people under 18, people with no age set and anyone logged out. |
NSFW_HIGH_PRECISION | Posts | X is fairly sure these posts contain adult content. Shown behind a warning, and hidden from the same people. |
NSFW_TEXT | Posts | X thought the words in these posts were sexually explicit. Hidden from the same people. |
NSFW_CARD_IMAGE | Posts | The picture on a link card looked like adult content. Behind a warning, hidden from strangers. |
NSFW_ADMIN | Posts | The posts came from an account X marked, after reports, as posting mainly adult or violent content. |
GORE_AND_VIOLENCE_HIGH_PRECISION | Posts | X is fairly sure these posts contain violent content. Behind a warning, and hidden from the same people. |
NsfwHighRecall, NsfwHighPrecision, NsfwNearPerfect | Account | X’s systems thought the account posts adult content. Its posts are hidden from strangers. |
NsfwAvatarImage, NsfwBannerImage | Account | X thought the profile photo or the header might be adult content. |
NsfwAdmin | Account | After reports, X marked the account as posting mainly adult or violent content. |
If you post this kind of content on purpose, X lets you mark your media as sensitive in your settings. If you don’t, the account labels often point to a profile photo or a header picture.
Abuse, hate and elections
The FOSNR labels keep a post findable only on your profile, with a label everyone can see. X applies most of them after people report a post.
| Label | In plain words |
|---|---|
FOSNR_ABUSE | After reports, X decided the post broke its abuse and harassment policy. |
FOSNR_HATEFUL_CONDUCT | X decided the post broke its hateful conduct policy. |
FOSNR_VIOLENT_SPEECH | After reports, X decided the post broke its violent content policy. |
FOSNR_CIVIC_INTEGRITY | After reports, X decided the post broke its civic integrity policy, about elections and voting. |
FOSNR_ABUSE_INSULTS | X decided the post insulted someone. It’s hidden from strangers, with a label everyone can see. |
X says authors can appeal a label that limits a post.
Links, reviews and emergencies
| Label | In plain words |
|---|---|
MALICIOUS_URL | X thought a link in the post leads to a harmful site. Hidden from strangers. |
DO_NOT_AMPLIFY | An older version of the harmful link label, which X is replacing. |
PDNA | X thought the post broke its Terms of Service, and holds it back while it reviews. |
BOUNCE | X decided the post broke its Terms of Service and asked you to delete it. Not shown meanwhile. |
FOR_EMERGENCY_USE_ONLY | An emergency label during an incident. The post sits behind a notice, out of Home. |
BRAZIL_ELECTION_LEGAL | AI-made media about Brazil’s elections during the silence period. Hidden in Brazil only. |
Account labels
| Label | In plain words | What to do |
|---|---|---|
ReadOnly | You could read posts but couldn’t post while a post that broke X’s Terms of Service stayed up. | Remove the post X points to, or appeal it. |
Compromised | X thought someone else got into the account. | Change your password and turn on two-factor login. |
AbusiveHighRecall | X thought the account might be automated or breaking its authenticity policy. | Complete the security check, and post by hand. |
ImpersonationHighPrecision | X thought the account was pretending to be someone else. | Make your name, photo and bio clearly your own. |
DoNotAmplify | X held back the account’s posts from strangers while it reviews. | Wait for the review, then appeal if it stays. |
Legal and copyright labels
LegalRequest, BystanderReport and UnspecifiedReason mean X withheld posts after a legal demand or a report under local law. The label names the countries in brackets, such as LegalRequest(DE), and the posts still show everywhere else. When the label covers every country, the posts aren’t shown on X. Dmca and is_dmca come from a copyright claim: the first takes a post down, the second removes its media and hides it from strangers.
What to do about a label
- Read the effect line. Hidden from strangers is a smaller problem than not shown on X.
- Look at the month. The report covers one month, so match it to what you posted then.
- Change what fits the label. For spam labels, that’s pace: spread your replies over the day, one reply per post, never the same words twice. Our guide to Show probable spam has the details.
- Appeal what X got wrong. X says authors can appeal a label on a post.
Twick’s Safe mode reminds you when you pass 30 replies in an hour, 100 in a day, 5 to the same person in a day, or two replies only seconds apart. It’s a reminder you can ignore, and no tool can promise X will never put a label on an account.
Questions
What does SPAM_HIGH_RECALL mean on X?
X’s spam filter thought some of your posts might be spam. Your followers still see them, but X stops recommending them to people who don’t follow you.
What does SpamHighRecall mean?
The account version of the same label. On the days it was on, X didn’t recommend any of the account’s posts to people who don’t follow it.
What does FOSNR mean in Under the Hood?
The FOSNR labels keep a post findable only on your profile, with a label everyone can see. X uses them for posts it decides broke its rules on abuse, hateful conduct, violent speech or civic integrity. The letters match the name of X’s 2023 policy, “Freedom of Speech, Not Reach”.
Which posts got the label?
The report doesn’t say. It counts how many of your posts got each label that month.
Can I see Under the Hood for someone else’s account?
No. The report only covers the account you’re logged in with.