From ‘Dictator Trump’ to ‘Rebel’: An Inside Look at Apple’s Confidential 100,000-Word List That Your iPhone Will Not Autocomplete

Try Our Free Tools!
Master the web with Free Tools that work as hard as you do. From Text Analysis to Website Management, we empower your digital journey with expert guidance and free, powerful tools.

According to a recent Reuters report, Apple curates an extensive multilingual catalog encompassing tens of thousands of words and phrases that its iPhone keyboard deliberately omits from predictive text, autocorrect, and swipe-typing functionalities. This information comes from an independent researcher.

This catalog, referred to as a digital ‘grey list’, reportedly consists of nearly 100,000 terms and their variations, though it does not hinder users from manually inputting them in messages, emails, or various applications.

This revelation prompts inquiries regarding the criteria Apple employs to determine which words warrant exclusion from keyboard suggestions and how it reconciles the necessity of censoring offensive language with the accurate prediction of user intent in typing.

Derik Kauffman, the researcher behind this discovery, noted that the majority of flagged words tend to be vulgar or offensive, often pertaining to drugs, sex, and violence.

Nonetheless, the list also encompasses politically sensitive expressions related to political figures, immigration, and ongoing territorial disputes.

Apple has yet to publicly acknowledge the existence or contents of this extensive list.

Politically Sensitive Phrases Excluded by Apple

Among the expressions in English that are reportedly omitted from predictive text are ‘dictator Trump’, ‘sleepy Joe’, ‘right wing conspiracy’, ‘Tibetan independence’, and ‘United Ireland’.

The catalog also includes phrases in various languages, with different entries flagged in accordance with the language utilized.

For instance, the Hindi keyboard reportedly restricts users from swipe-typing ‘Modi braggart’ or ‘Rahul is Pappu’, both of which are derogatory comments aimed at Prime Minister Narendra Modi and opposition leader Rahul Gandhi, respectively.

Further examples mentioned by Kauffman include the French phrase ‘immigration too high’, which French speakers are unable to swipe-type, and ‘occupied territories’, flagged in Hebrew.

“Every major language exhibits some variation of this phenomenon,” Kauffman remarked.

In a statement to Reuters, Apple asserted that users have consistently been able to manually type specific words and phrases, thereby circumventing predictive text suggestions.

The company emphasized that its keyboard aims to eschew inappropriate or offensive language suggestions.

“This represents a very subtle form of censorship,” noted Kauffman. “Yet, at the magnitude that Apple operates, even slight restrictions can yield significant outcomes.”

Researcher Discovers Inconsistencies in Apple’s Filtering

Kauffman expressed an understanding of Apple’s goal to shield users from inadvertently disseminating obscene or incendiary messages. Nonetheless, he raised concerns regarding the uniformity in the application of the company’s filtering system.

For instance, he found that users could swipe-type ‘read smut’, while the keyboard obstructed ‘read gay smut’.

Additionally, ‘bad boy’ and ‘white man’ reportedly posed challenges for swipe-typing, whereas ‘bad girl’ and ‘black man’ proceeded without such impediments.

These disparities prompt inquiries into the criteria that Apple employs to decide which expressions to exclude from its keyboard suggestions.

Discovery of Apple’s Obscured Word List

Kauffman recounted that he stumbled upon the list in September after becoming aware of unusual disruptions while utilizing the swipe-typing feature on the iPhone.

The swipe-typing function enables users to input words by gliding their fingers continuously over the keyboard instead of tapping individual letters.

He observed that certain combinations, notably ‘bad boy’ and ‘white man’, posed challenges, prompting further investigation into the cause.

Subsequently, he employed OpenAI’s Codex and Anthropic’s Claude Code to analyze the operating system and pinpoint the root of the issue.

A smartphone displaying the word Anthropic lies on a wooden desk near a mug and two potted plants.

Ultimately, his examination unveiled a trove of restricted expressions intricately embedded within Apple’s operating framework.

Kauffman approximated that the list encompasses roughly 100,000 words and phrases, including their numerous variations, although this estimate excludes Chinese characters.

Reuters Conducts Examination of Apple’s iPhone Keyboard

Reuters independently assessed a select array of the expressions identified by Kauffman and discerned that they were challenging to enter through swipe-typing or were absent from autocomplete suggestions.

For instance, attempts to swipe-type ‘United Ireland’ frequently resulted in ‘United Iceland’, while ‘Tibetan independence’ converted into ‘Tibetan independences’.

Likewise, ‘white man’ occasionally morphed into ‘white mean’, and ‘sleepy Joe’ produced ‘Sleepy hope’.

While the phrase ‘dictator Trump’ could technically be input via swipe-typing, it was not suggested through autocomplete; instead, the keyboard proffered ‘dictator trumps’.

Source link: Livemint.com.

Disclosure: This article is for general information only and is based on publicly available sources. We aim for accuracy but can't guarantee it. The views expressed are the author's and may not reflect those of the publication. Some content was created with help from AI and reviewed by a human for clarity and accuracy. We value transparency and encourage readers to verify important details. This article may include affiliate links. If you buy something through them, we may earn a small commission — at no extra cost to you. All information is carefully selected and reviewed to ensure it's helpful and trustworthy.

Reported By

Neil Hemmings

I'm Neil Hemmings from Anaheim, CA, with an Associate of Science in Computer Science from Diablo Valley College. As Senior Tech Associate and Content Manager at RS Web Solutions, I write about AI, gadgets, cybersecurity, and apps – sharing hands-on reviews, tutorials, and practical tech insights.
Share the Love
Related News Worth Reading