GoKawiilResearcher reverse-engineers Apple's hidden profanity-tagging system
A developer investigated macOS's ~/Library/Keyboard/VulgarWordUsage.db, which logs swear-word usage per app and recipient to help autocorrect adapt, and found it empty on their own Mac. Digging further, they used undocumented private functions from Apple's Lexicon.framework to extract the underlying system dictionary, which tags every word with a vulgarity tier (vulgar, offensive, sensitive) and up to eleven topic categories like sex, drugs, religion and politics, publishing full word dumps for 15 languages.
GoKawiil's interpretation of the reporting above, not reported fact.
The findings suggest Apple maintains extensive, undocumented linguistic classification systems that shape autocorrect and typing behavior without any public API or transparency for users or developers. The reported quirks—such as 'Trump' being the only living politician tagged sensitive alongside historical dictators, while other world leaders are unflagged—raise questions about how these classifications were compiled and whether they reflect consistent or arbitrary editorial choices by Apple.
- Apple's VulgarWordUsage.db tracks swear-word usage per app and recipient but was found empty on the researcher's device.
- Using private Lexicon.framework functions, the researcher extracted Apple's hidden word-tagging system with vulgarity tiers and topic flags.
- Notable anomalies include 'Trump' being tagged sensitive alongside dictators, while other current world leaders are untagged.
Source: helgesver.re, 2026-11-03
Published there as: “I Dumped Apple's Fucking Word List”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.