Apple Maintains Secret List of Banned Autocomplete Phrases
Left 50%
Center 0%
Right 50%
1 left · 0 center · 1 right
What happened
On November 6, 2026, Reuters first reported that researcher Derik Kauffman had discovered a previously undisclosed multilingual list of about 100,000 words and phrases that Apple’s iOS keyboard will not offer as autocomplete suggestions or accept via swipe-typing, though users can still enter them letter by letter; The Verge credited Reuters as the origin, with Kauffman and Reuters as the first on-record sources. Kauffman located the list stored as a hash table inside iOS using OpenAI’s Codex and Anthropic’s Claude Code and published it; engineer Matt Suiche independently recovered a near-perfect match, and Reuters found that a sample of the phrases were difficult to swipe-type—often yielding altered text—or were not suggested. Most entries are vulgar, offensive, or related to sex, drugs, or violence, while others are politically or otherwise sensitive, including phrases such as “dictator Trump” and “sleepy Joe”; Apple said its keyboard is designed to avoid potentially inappropriate or offensive suggestions and that users can type what they want or override predictions, but did not comment directly on the list’s size, composition, or reasons for flagging specific political terms.
Sourcing check · Fully on the record · 2 articles
2 of 2 articles cite a named source for the story. It was on the record from the first report, credited to Reuters. Most articles credit Reuters rather than reporting it themselves.
What these terms mean and how they're counted
- On the record · names a source 2 of 2
- On the record means the article cites a source by name for the story's main claim: a person, or an institution speaking officially, such as a company statement or a named agency's spokesperson. The opposite is an unnamed source. The count is how many of the counted articles cite at least one such source; the names are everyone named across them, each once however the articles spell it. Here: Derik Kauffman, Reuters and Apple
- Primary evidence · cites a document or data 0 of 2
- Primary evidence means the article cites a record you could check yourself: a court filing or ruling, a dataset or report, a recording or transcript, an official statement or press release, market data. A quote is not a document. The count is how many articles cite one. Here: None cited.
- Credits another outlet 1 of 2
- The article says the story comes from another outlet's reporting (“first reported by Reuters”). Those articles relay the story; they are not independent confirmation. Here: Reuters
- Unnamed sources 0 of 2
- The main claim is attributed to people who are not identified: “people familiar with the matter”, “officials said”. The count is how many articles do this.
- How it developed
- When the first report landed and whether it named anyone; when the first named source and the first document appeared. Here: First report Nov 6, 15:10 UTC by The Verge, crediting Reuters, naming its source.
- Not counted 0
- Opinion, analysis and video pieces, and articles too short to assess. Every count above is out of the counted articles. An article republished under a new headline is one article.
This is about who is cited, not whether the story is true. Each source below says what it names. How we count
Omitted — what each side leaves out
Unpacked
Each side withheld a different slice of Apple's political blocklist, and the withheld slice changes what the filter appears to target. Left-leaning coverage in The Verge includes a covid-hoax phrase and illegal-immigrant language that right-leaning coverage never names. Right-leaning coverage in the New York Post instead reports "Tibetan independence," which left-leaning coverage omits entirely. A reader of only one side would not see the domestic culture-war terms and the foreign separatist slogans together.
Right-leaning coverage foregrounds a researcher's line that the feature is "a very light-touch form of censorship" and later cites a lawyer who wanted the list out in the open. Left-leaning coverage never introduces those voices and stays with the contents of the list plus Apple's product explanation.
Who inside Apple decides that a separatist slogan belongs on the same list as a slur, and under what written policy?
Bottom line
You cannot reconstruct Apple's political filter from one paper: The Verge ran the covid-hoax phrase the New York Post skipped, and the Post ran separatist slogans The Verge never printed.
The receipts — 5 facts one side's coverage lacks
- The iOS gray list includes the phrase Tibetan independence
- Swipe-typing can change left wing conspiracy to left wrong conspiracy
- A conservative count of gray-listed phrases is about 100,000, not including Chinese characters
- The phrase list was stored as a hash table that was converted back to text
- Attempts to swipe-type white man yielded white mean
The Left View
Left-leaning coverage treated the find as a “secret list of banned autocomplete phrases,” stressing that swipe-typing can automatically rewrite listed terms—The Verge cited “left wing conspiracy” becoming “left wrong conspiracy”—and that the roster mixes slurs and sex- and drug-related language with political items including “illegal immigrant,” “covid was a hoax,” and “left wing conspiracy,” plus oddities such as “Santa isn’t real.” It also highlighted uneven application, such as blocking “bad girl” but not “bad boy,” and noted that Apple did not immediately respond to The Verge’s request for comment.
The Right View
Right-leaning coverage, while stating “the list is not a ban,” framed it as what Kauffman called “a very light-touch form of censorship” that at Apple’s scale “only takes a light touch to have a big impact” and could “quietly shape how they communicate.” The New York Post foregrounded additional politically charged English and non-English entries—“right wing conspiracy,” “Tibetan independence,” “United Ireland,” and phrases about immigration and occupied territories—plus uneven treatment such as “white man” versus “black man” and “read gay smut” versus “read smut,” and quoted Electronic Frontier Foundation counsel David Greene that the list is “creepy” and “should not be secret,” with users needing awareness and an ability to customize.
Our Take (balanced)
The left-leaning account’s strongest point is that this is a suggestion filter, not a bar on typing, and that political examples run in more than one direction—both anti-Trump and anti-Biden or COVID-skeptic phrases appear—while Apple’s stated aim, and the bulk of the list, is avoiding vulgar or illicit predictions. The right-leaning account’s strongest point is that secrecy plus scale can still steer everyday wording, backed by Kauffman’s “light-touch form of censorship” characterization, documented swipe-type substitutions, geographically sensitive phrases, and Greene’s argument that users should know and be able to customize. The central unresolved tension is whether an undisclosed, non-customizable keyboard filter that leaves letter-by-letter typing intact is best understood as a product-safety choice or as unaccountable influence over how hundreds of millions of people phrase messages.
2 sources
More stories
The week's bottom lines, in your inbox
One email a week: the five stories that mattered and what they actually mean. Free.