X Says My Content Is Sensitive, but It Isn't (October 2026)
X's code, read October 16, 2026: labels from your profile picture, banner, link previews and text, and a rule that labels the whole account for 7 days.

X can mark your account as sensitive even if you never post adult content. Its labels come from automated systems that read more than your posts: your profile picture, your banner, the image in a link preview and the words you use. X's code, read on October 16, 2026, also has a rule that labels the whole account for 7 days when 3 of your last 5 posts are labeled.
A sensitive label hides your posts from people who do not follow you. You still see everything as usual, which is why it often goes unnoticed.
Where do X's sensitive labels come from?
X's labels are applied by automated systems, sometimes after a user report. X's Under the Hood tool explains each label it reports. These are the sensitive ones in the published code:
| Label | What X says triggers it |
|---|---|
NSFW_HIGH_RECALL |
A post that "may contain" content subject to the Adult Content policy |
NSFW_HIGH_PRECISION |
A post "likely to contain" such content, by automated systems or a report |
NSFW_TEXT |
A post likely to contain "adult sexually explicit language" |
NSFW_CARD_IMAGE |
A post whose link preview image is likely adult content |
NsfwAvatarImage |
An account flagged as possibly posting adult content |
NsfwBannerImage |
An account flagged as possibly posting adult content |
The two account labels are named after the profile picture and the banner. X's Adult Content policy forbids adult content in "profile pictures, headers, List banners, or Community cover photos". A borderline avatar can therefore label the account, whatever you post.
The words "may contain" matter. NSFW_HIGH_RECALL is the broad label: a high-recall system is built to catch as much as possible, so it also catches posts that are not adult.
Can a few posts get the whole account labeled?
Free, no login, results in seconds.
Yes. X's code has a rule that turns post labels into an account label. With the default published in the code, if 3 of your last 5 posts from the past 60 days carry NSFW_HIGH_PRECISION, the account gets the same label for 7 days.
The rule skips an account that already has the label, so it is not stacked while it lasts. A second default rule, for the label POSSIBLY_NSFW_ACCOUNT, asks for 11 matching posts out of the last 10, so it cannot trigger with the default values. X can replace both rules through a feature switch without changing the code.
Who stops seeing your posts?
The answer depends on who is looking. In X's published visibility rules:
- People who do not follow you lose your labeled posts from recommendations. Under the Hood describes every sensitive account label above as "The account's posts are hidden from recommendations to non-followers."
- Logged-out visitors and viewers under 18 do not get posts with sensitive media labels, even from accounts they follow.
- In a list of countries set by X, viewers with no stated age do not get them either.
- Followers who have not allowed sensitive media see the media behind a blur.
X's Adult Content policy confirms the age part: "Users under 18 or viewers who do not include a birth date on their profile cannot click to view marked content."
Can X turn on the sensitive media setting for you?
Yes. X's Adult Content policy (May 2024) asks accounts that regularly post adult content to adjust their media settings, which puts all their images and videos behind a content warning. It adds: "If you continue to fail marking your posts, we will adjust your account settings for you."
Check the setting yourself in Settings and privacy, Privacy and safety, Your posts: "Mark media you post as having material that may be sensitive". If you do not post adult content and it is on, that is worth an appeal.
What does X not count as adult content?
X's Adult Content policy lists content that "is not considered Adult Content and does not need a content warning":
- educational content, such as breast cancer awareness;
- medical content, such as post-surgery photos;
- partial or digitally obscured nudity, such as modeling or beach photos;
- activist campaigns carried out with nudity;
- artistic nudity, including paintings, drawings, sculptures and statues;
- parody or satire with a political point, such as caricatures;
- humor with no intent to cause sexual arousal, such as memes.
An automated system can still label this kind of content. The policy is what you appeal with.
What should you do if X labels you by mistake?
These steps follow from X's code and policy.
- Look at your profile picture and banner first. They carry their own account labels, and X forbids adult content there.
- Check the links you share. A link preview image can label a post that has no media of its own.
- Check your media setting, and leave it off if you do not post adult content.
- Appeal. The policy says: "If you believe we made a mistake, you can submit an appeal."
- Open X's Under the Hood if you have access. X calls it a pilot, and it shows some labels applied over the past month. Our page on X's Under the Hood explains what it shows and what it does not.
Avoid three posts in a row that could be read as adult. With the default rule, 3 labeled posts out of your last 5 label the whole account for 7 days.
How do you know whether your account is restricted?
A label explains a cause, not what other people see. To see the effect, check whether X restricts your account. A free check on Shadowban Radar tests the 4 restrictions on any public handle, with no login: search suggestion ban, search ban, ghost ban and reply deboosting. The search suggestion ban page explains the first one, and our glossary defines each term.
Sources
Read on October 16, 2026, in the xai-org/x-algorithm repository at commit 92f0918, published October 15, 2026, and on X's Help Center.
- underTheHoodLabels.strato: what each sensitive post label means and does.
- underTheHoodLabels.strato: the sensitive account labels, including profile picture and banner.
- postToUserLabelRules.strato: the rule that labels an account from its posts, and its default values.
- postToUserLabelRules.strato: how matching posts are counted.
- author_rules.rs: the account labels that hide posts from other people.
- tweet_rules.rs: posts hidden from logged-out, underage and no-stated-age viewers.
- levels.rs: which rules apply to followed accounts and to recommendations.
- X Help Center, Adult Content policy (May 2024): what counts, highly visible areas, settings changed by X and appeals.
Shadowban Radar is not affiliated with X Corp. For the rules that shape reach on every account, see what X's code says about reach.


