Why a keyboard can know a swear word and still avoid suggesting it
A phone keyboard does not rely on one simple list of words it either knows or does not know. It estimates what you are trying to type from the letters on screen, the words around them, and patterns learned from language. That lets it recognize an offensive word well enough to correct a typo or stop changing it—without offering that word as a suggestion.
The distinction is between recognition and recommendation. A keyboard may treat a term as valid input but leave it out of the prominent suggestions shown above the keys. That can reduce the chance that an accidental tap or a shared-screen moment puts an unwanted word into a message. If you deliberately type the word, the keyboard may accept it, even if it did not volunteer it.
The filter is part of the product’s design
Suggested text is a judgment about what to display, not a neutral transcript of every word the software can process. Keyboard makers can tune suggestions to avoid language they consider inappropriate, while still allowing users to write it. The exact balance—and whether a word is corrected, retained, or suggested—can vary with the keyboard, its settings, and its language model.
That is why a keyboard’s apparent “censorship” can seem inconsistent. It may silently steer a mistyped word toward a cleaner one, yet leave the same word alone when the spelling is clear. The phone is not offended; it is applying separate rules for understanding what you typed and deciding what to recommend next.
