Struggling to spell ‘male chauvinist pig’? Apple’s grey list means your iPhone won’t help you

by | Nov 11, 2026 | Technology

Struggling to spell ‘male chauvinist pig’? Apple’s grey list means your iPhone won’t help you

Research conducted by independent investigator Derik Kauffman has uncovered evidence that Apple maintains an extensive list of restricted words and phrases within its iOS operating system. The list, which contains more than 100,000 entries, prevents Apple’s predictive text feature from suggesting or autocompleting these terms when users attempt to type them.

The restricted words span multiple categories. Some entries are relatively expected, including references to controlled substances, explicit sexual content, and phrases like “iMessage isn’t working.” However, the list also includes terms that Kauffman characterized as potentially politically sensitive. Examples include “Trayvon Martin,” “Kashmir dispute,” “Muslim extremist,” and “male chauvinist pig.” The restrictions extend to multiple languages as well, with Apple blocking autocomplete for phrases such as “the occupied territories” in Hebrew and “the Hamas movement” in Arabic.

When users attempt to swipe-type these restricted phrases, the autocomplete feature often produces unexpected results or fails to complete the intended text. For instance, attempts to type “United Ireland” may instead autocomplete to “United Iceland,” while “Tibetan independence” may become “Tibetan independences.”

Apple responded to the findings by stating that its keyboard is designed to prevent inappropriate or offensive suggestions. The company noted that users retain the ability to manually type any words or phrases they wish and can override predictive text suggestions to achieve their desired results.

The discovery provides insight into how autocomplete technology functions behind the scenes. Apple’s system has evolved significantly over time, from basic word-matching algorithms to more sophisticated n-gram statistical models. More recently, the company introduced a transformer-based autocorrect system comparable to large language models, which some users noticed produced unusual spelling suggestions during earlier deployments.

Article Attribution | Read More at Article Source

Article summary produced by Claude AI