The word covers two very different claims, and separating them makes the topic tractable.
What is documented
Platforms reduce the distribution of content that sits near a policy line without crossing it. Meta has described this publicly as reducing recommendations for borderline content. Content in restricted hashtags is filtered. Accounts with recent violations have their reach limited for a period. All of this is real, and none of it comes with a notification.
What is not documented
The stronger version — that platforms silently blacklist ordinary accounts for arbitrary reasons, or for using too many hashtags, or for posting too often — has no evidence behind it. Every time a viral "shadowban list" circulates, it turns out to describe normal variance.
How to check
Post, wait thirty minutes, then search a small tag you used from a logged-out browser or a second account. If your post appears in the recent section of small tags, you are being distributed normally.
Also check your account status page. Instagram, TikTok and YouTube all provide one, and it names actual restrictions if any exist.
The most useful reframe
Distribution is not a right that gets taken away — it is allocated post by post based on predicted response. A post that gets less of it did not get punished, it got predicted less confidently. That framing is less satisfying and considerably more actionable.