Researcher Discovers Apple’s iOS Keyboard Filters Specific Terms

Apple has been quietly omitting a vast array of words and expressions from the autocomplete and swipe‑type features of its iPhone keyboard. The hidden list, said to contain more than 100,000 entries across dozens of languages, includes politically loaded phrases such as "dictator Trump" and conspiracy‑themed statements like "covid is a hoax." Independent researcher Derik Kauffman uncovered the practice, noting that users can still type the barred words manually, but the predictive suggestions that normally speed up typing will not appear for those terms.

Kauffman reached his conclusion by methodically testing the iOS keyboard’s suggestion engine, logging which strings showed up after each keystroke. His investigation exposed a sizable blocklist operating behind the scenes, preventing certain phrases from showing in the predictive row above the keyboard. He points out that the filter runs across "every major language" supported by iOS, implying a centrally administered policy rather than isolated, language‑specific adjustments.

The way the blocklist works is simple: when a user begins entering a prohibited phrase, the keyboard simply leaves it out of the autocomplete options, even though the underlying dictionary still holds the entry. Consequently, typing the phrase one character at a time remains possible, but the convenience of predictive text is lost. Apple has offered no public comment on the list’s existence, and its official documentation makes no mention of content‑based filtering within the keyboard software.

This finding adds another note to the growing chorus of criticism aimed at Apple’s opaque content‑moderation practices. While the company has already faced scrutiny over app‑store policies and the removal of certain applications, the discovery that its own input method can silently shape the language users encounter highlights a less visible layer of control. Critics argue that any undisclosed algorithmic censorship can undermine user trust and raise questions about the criteria used to deem phrases unacceptable.

Legal analysts warn that the secretive nature of the blocklist could draw regulatory scrutiny, especially in regions tightening rules on algorithmic transparency. Consumer‑privacy advocates are urging Apple to reveal the scope and reasoning behind the filtered terms, asserting that users deserve to know when technology is influencing the words suggested to them. As the story unfolds, observers will be watching for Apple’s response, possible updates to its privacy policies, or additional research that might illuminate how widespread such filtering is across other Apple services.