In recent months, several users of Apple devices have complained on social networks about the ineffectiveness of the brand’s artificial intelligence, marketed as Apple Intelligence, in distinguishing between fraudulent and real communications. Apple’s AI, which among its main functions for iPhone or Mac includes summarizing emails and SMS and marking messages as priority, does not seem to distinguish at all between fraudulent emails and really important information, which can pose an added risk for users. .
“Maybe Apple Intelligence shouldn’t mark fraudulent emails as “Priority” with a summary saying they are for security reasons?,” comments @howard.fm wryly on Bluesky. The message summary feature is one of the key features of Apple Intelligence.
One of the examples that Apple gives of the usefulness of this tool is to summarize a conversation about a meeting with friends in a phrase like “brunch after football on Saturday; restaurant or host at home is suggested”, which informs the user in a few words.
Be alert to scams
However, if Apple Intelligence is not able to distinguish between real communications and scam attempts, some users could trust themselves and not notice other signals before falling for a scam. In the example given by the user, Apple Intelligence marks as a priority a phishing email that impersonates an entity that requests data management consent.
Similarly, user @colin.hoagie.fun notes on the same social network: “Can you imagine what would happen if Apple Intelligence actually flagged fraudulent messages instead of summarizing them in a useful way?” The same thing happened to this user, but with an SMS message instead of an email. In this case, the SMS impersonates the identity of the USPS package delivery company.
Instead of detecting signs of fraud in this message, Apple Intelligence sends a notification with a summary, asking the user to update the information in less than 24 hours in order to receive the supposed delivery. Additionally, “AI rewriting helpfully removes all grammatical cues that indicate this could be fraud,” as another user notes.
Thus, we see that scammers and cybercriminals can not only use rewriting tools like ChatGPT to improve their phishing emails, but tools like Apple Intelligence can also collaborate with them inadvertently, by sneaking these types of messages without distinction. In recent weeks, not only has this issue been a controversy about Apple’s AI; It has also been in the news due to its treatment of fake news.
Apple Intelligence also sneaks in fake news
Last month, the British media BBC published an article in which it warned that Apple Intelligence was generating false summaries of current headlines, causing the spread of fake news, in addition to damaging the reputation of the BBC by making it believe that this media outlet is the that issued the information. The BBC collects several cases, including a notification in which the AI summary indicates that Luigi Mangione, the young man arrested for the murder of Brian Thompson in the US, has shot himself, and in another, indicates that the tennis player Rafa Nadal has made some comments about his sexuality. Both informations are false.
The same has happened with the New York Times: Apple Intelligence released a notification saying that Israeli Prime Minister Benjamin Netanyahu had been detained. For these reasons, the British journalists’ association NUJ has asked Apple to remove this function from its devices until the problem is corrected.
With all this, users see the seams and weaknesses of Apple Intelligence. Despite Apple’s caution when launching its own AI tool, the time delay with which it was launched compared to competitors such as ChatGPT has not helped them to fully polish aspects like these.
