Apple introduced users to its Apple Intelligence features with the recent release of the iOS 18.1 developer beta. Among the new services in this update is Writing Tools, which leverages Apple’s AI models to reformat and rewrite text. However, the tool issues a message cautioning that AI-generated suggestions may not always be of the highest quality, especially when dealing with certain topics.
Apple Intelligence can be accessed across the system to assist with text revision. But attempting to rewrite content containing swear words like “sh**” and “bastard” triggers a warning indicating that Writing Tools is not equipped to handle such content, with a notice stating that the quality of revisions may vary.
This limitation has been noted by other users, with instances where the rewritten text fails to maintain the original tone.
The warning also appears when writing about topics such as drugs, killing, or murder.
Despite the warning, Apple Intelligence still provides suggestions for sentences containing words or phrases that it hasn’t been trained on. In a test scenario, replacing “sh***y” with “crappy” led to the same suggestion from the AI tool.
We have reached out to Apple for further clarification on the topics Writing Tools are not trained to provide suggestions on, and will update the story with any response from the company.
Apple may be taking a cautious approach to avoid controversy by restricting certain words, topics, and tones when its AI rewrites sentences. The focus of Apple Intelligence-powered writing tools is not to generate new content entirely, but rather to assist users with text revision. This could be a preventative measure to avoid any backlash from regulators due to potentially problematic content being generated.
It took Apple years to remove a block on swear words from the keyboard’s autocorrect suggestions, finally offering a feature in iOS 17 that learns the user’s preferred swears. Perhaps with Apple Intelligence, Apple is simply exercising caution to prevent any accusations of generating inappropriate content.
