A counter point to this adage in modern times is that censorship seems to spread as a result of users sharing content across platforms with varying levels of moderation. I've seen many examples of "shorts" being shared on FB or Instagram which originated from TikTok and which feature heavy use of either euphemism (eg. "unalived" instead of "murdered") or even explicitly silenced language.

Platforms which do not heavily moderate content will nonetheless still have heavily self-censored content as a result of users being conditioned by other platforms into self-censorship.

I can confirm that content censored to suit restrictive platforms finds it's way to less restrictive platforms (e.g. a screenshot of a Tumblr post censored for Meta may find it's way to Imgur)

I have noticed that some users on Imgur will make an effort to de-censor content though (e.g. re-adding the censored text in a screenshot)