Apple’s grey list: Words and phrases your iPhone won’t help you spell

Apple’s iPhone keyboard is reported to use a hidden “grey list” of words and phrases it will not autocomplete or swipe-type, ranging from profanity and sexual content to political expressions like “Tibetan independence,” “United Ireland,” and “Trayvon Martin.” Commenters are split between seeing this as a reasonable safeguard against embarrassing or dangerous mis-suggestions and as a troubling form of opaque, soft censorship that subtly shapes what billions of users find easy or hard to express. Many also complain that Apple’s keyboard has become less accurate and more intrusive over time, and argue for transparency, user controls, or on-device, user-specific prediction instead of centrally curated blocklists.

Scope of Apple’s “grey list” behavior

  • iOS won’t autocomplete or swipe‑type certain words/phrases, but you can still peck them out manually.
  • A community project (swipeblocked.com) catalogues these phrases and shows they affect swipe input, next‑word prediction, and sometimes corrections.
  • Examples range from profanity and sexual/drug terms to political phrases (“Tibetan independence”, “United Ireland”), names (“Trayvon Martin”), and even “iMessage isn’t working”.
  • Some combinations trigger bizarre rewrites (e.g., “left wing propaganda” → “left wring propaganda”; “is important” → “is impotent”; “black woman” → “blank woman”).

Censorship vs. user‑protection

  • One camp calls this censorship or “soft coercion”: the device subtly steers language by making certain views harder to express, especially when phrases are clearly intended.
  • Others argue it’s a safety/UX feature: swipe and prediction are probabilistic, and avoiding “high‑risk” phrases reduces accidental insults, slurs, or politically explosive text being sent to bosses, family, or in repressive regimes.
  • Some note there’s no outright ban—you can still type anything—so they see it as Apple censoring its own suggestions, not users’ speech.

Bias, politics, and arbitrariness

  • Commenters highlight asymmetries: you can call some leaders “dictator” but not others; certain countries can be “bad/poor” but others can’t; only one full personal name is blocked.
  • COVID‑skeptic phrases reportedly appeared later, reinforcing a sense of shifting, opaque editorial lines.
  • Many see this as PR‑driven risk management and geopolitical appeasement (e.g., China‑sensitive terms), not a coherent ethical policy.

Technical behavior and keyboard quality

  • Multiple users say this goes beyond “not suggesting”: iOS rewrites earlier words to avoid forming blocked phrases and withholds the correct word from the suggestion bar.
  • Others point out all swipe and “fat‑finger” correction is inherently predictive guesswork; Apple is choosing safer guesses.
  • Independent of the grey list, many complain the iOS keyboard has regressed: over‑aggressive autocorrect, retroactive word changes, dynamic hitboxes, and worse performance than Gboard/Android.

Transparency, control, and alternatives

  • Many want:
    • A visible, versioned list of blocked phrases.
    • Toggles for: no autocomplete, filter on/off, or user‑only training.
    • Editable user wordlists.
  • Some use or recommend third‑party keyboards; others note those can have their own filters or privacy tradeoffs.
  • Broader concern: this is one more example of big platforms nudging language and behavior via defaults and opaque algorithms.