The BBC has complained to Apple after the tech giant’s new iPhone feature generated a fake headline about a high-profile murder in the US.
Apple Intelligence, launched in the UK earlier this weekuses artificial intelligence (AI) to summarize and group notifications.
This week, the AI-powered summary falsely made it appear that BBC News had published an article claiming that Luigi Mangione, the man arrested after the murder of health insurance CEO Brian Thompson in New York, had committed suicide . He didn’t do it.
A BBC spokesperson said the corporation had contacted Apple “to raise this concern and resolve the issue.”
Apple declined to comment.
“BBC News is the world’s most trusted news media,” added the BBC spokesperson.
“It is essential to us that our audiences can trust any information or journalism published on our behalf, including notifications.”
The notice that contained false statements about Mangione was otherwise accurate in its summaries about the toppling of Bashar al-Assad’s regime in Syria and in an update on South Korean President Yoon Suk Yeol.
But the BBC does not appear to be the only news publisher whose headlines have been distorted by Apple’s new AI technology.
On November 21, three New York Times articles on different topics were consolidated into a single notification, part of which read “Netanyahu Arrested,” referring to the Israeli prime minister.
It inaccurately summarized a newspaper report that the International Criminal Court had issued an arrest warrant for Netanyahu, rather than any reporting on his arrest.
The error was highlighted on Bluesky by a journalist from the American investigative journalism site ProPublica.
The BBC was unable to independently verify the screenshot and The New York Times declined to comment on BBC News.
“Embarrassing” mistake
Apple says one of the reasons people might like its AI-powered notification summaries is to help reduce interruptions caused by ongoing notifications and allow the user to prioritize notifications that are the most important. more important.
It’s only available on select iPhones – those using iOS system version 18.1 or later on recent devices (all iPhone 16 phones, the 15 Pro, and the 15 Pro Max). It is also available on some iPads and Macs.
Professor Petros Iosifidis, professor of media policy at City University London, told BBC News that Apple’s mistake “seems embarrassing”.
“I can see the pressure being first on the market, but I’m surprised Apple put its name on such an obviously half-baked product,” he said.
“Yes, the potential benefits exist, but the technology is not there yet and there is a real danger of misinformation spreading.”
Bulk notifications are marked with a specific icon and users can report any concerns they have on a summary of notifications on their devices. Apple did not say how many reports it received.
Apple Intelligence doesn’t just summarize publishers’ articles, and it has reported summaries of emails and text messages sometimes I don’t quite reach the goal.
And this isn’t the first time a major tech company has discovered that AI summaries don’t always work.
In May, in what Google described as “isolated examples,” its AI Overviews tool for Internet searches told some users looking for how to stick cheese to pizza that they should consider using “glue.” non-toxic.
The search engine’s AI-generated responses also indicated geologists recommend humans eat a stone a day.