Apple Intelligence's writing tools stumble on swears and controversial topics

Apple gave users the first real taste of its Apple Intelligence features with the release of the iOS 18.1 developer beta last month. Writing Tools, which lets you reformat or rewrite text using Apple’s AI models, is one of the new always-available services in this update. But the tool will show you a message notifying you that AI-generated suggestions might not be of the highest quality when tackling certain topics.

You can call up Apple Intelligence pretty much anywhere across the system when you need to rejig text. However, when you try to rewrite a paragraph or a sentence that has swear words, including “s—” and “bastard,” you will see a warning saying, “Writing Tools was not designed to handle this type of content,” with a header text warning that the quality of rewriting may vary.

Swear words are not the only things that trigger this warning. Mentions of drugs, killing, or murder will also bring up this warning.

Even when Apple Intelligence shows the warning, it still offers suggestions for sentences when you use one of the words or phrases that it hasn’t been trained on. One time in testing, I replaced “sh—y” with “crappy” to make the warning go away, and the AI tool returned the exact same suggestion as before.

We have asked Apple to provide more information on what topics the writing tools are not trained to give a suggestions on. We will update the story if we hear back from the company.

Apple is likely avoiding any controversy by prohibiting the tackling of certain words, topics and tones when its AI is rewriting sentences. Apple Intelligence-powered writing tools are not geared to generate new content from scratch, anyway. However, Apple wants to warn people when AI is looking at these terms.

It took Apple years to remove a block on swear words from the keyboard’s autocorrect suggestions. Last year, with iOS 17, Apple finally rolled out an autocorrect feature that learns your swears. Perhaps with Apple Intelligence, Apple is just being cautious to prevent regulators from blaming it for generating problematic content.