The Apple Intelligence features rolled out as part of iOS 18 are cool on paper. Like all generative AI, though, it’s not perfect—especially not at first. Apple’s AI summaries have produced weird results, and it seems like the issues are so widespread Apple is actually just choosing to pull them for now.
Apple is temporarily scaling back its Apple Intelligence notification summaries after a series of widespread issues. The feature, which uses AI to provide concise summaries of incoming notifications, has been plagued by accuracy issues, particularly when summarizing news and entertainment content.
The botched summaries are funny at best and actually harmful or misinformative at worst. Apple Intelligence seems to struggle to accurately capture the nuance of human language, and this has led to several instances where news summaries generated by the AI were factually incorrect or misleading. News outlets such as BBC News have voiced concerns that inaccurate summaries could damage the credibility of news organizations and erode public trust. If you look at your phone’s notifications and see an AI making up fake news, you won’t know immediately if something is fake/inaccurate unless you think of opening the post.
When looking at a notification at a first glance, you might not even think that it’s an AI summary at first either. Apple’s logic behind Apple Intelligence features is to make AI feel as native as it can be within your phone, but that becomes a problem if news summaries are making up their own news.
Apple has disabled news and entertainment notification summaries in the latest beta releases of iOS 18.3, iPadOS 18.3, and macOS Sequoia 15.3, and they will eventually be gone once the stable versions of these operating systems roll out within the next few days or weeks. The company is also working to improve the accuracy and reliability of the summaries and intends to reintroduce the feature in a future update once these issues have been addressed.
The company previously announced its intention to slap a warning label on its AI summaries, but the move was seen by many as Apple just putting a band-aid on the issue rather than properly solving it. The subsequent backlash might’ve prompted the company to actually take more concrete action on the issue and actually take time to work on it.
Another good part is that Apple is also adding some additional measures for transparency’s sake that aren’t just a lazy label that most people won’t read. For the content categories that will still be live following this partial removal, Apple is implementing several changes to enhance transparency and user control. When you enable notification summaries, you will now be explicitly informed that the feature is in beta and may produce unexpected results. Apple has also tweaked the visual style of summarized notifications, using italicized text to distinguish them from regular notifications. If you see italics, you might want to double-check whether the text in the notification is accurate.
Furthermore, Apple is giving users more granular control over notification summaries. You will now be able to toggle the feature on or off directly from the Lock Screen, and settings for summaries will be customizable on a per-app basis.
Hopefully, by the time this is reintroduced, summaries will be much more accurate. It also serves as yet another reminder that despite what some companies might tell you about how close we are to AGI, generative AI still produces slop sometimes and it’s way more common than you think. While we might eventually reach the point where it’s fine, even the most advanced models are still a work in progress and they might make mistakes sometimes.
Source: AppleInsider, MacRumors, 9to5Mac
