Apple has killed its Apple Intelligence AI news feature after it fabricated stories and twisted real headlines into fiction.
Apple’s AI news was supposed to make life easier by summing up news alerts from multiple sources. Instead, it created chaos by pushing out fake news, often under trusted media brands.
Here’s where it all went wrong:
- Using the BBC’s logo, it invented a story claiming tennis star Rafael Nadal had come out as gay, completely misunderstanding a story about a Brazilian player.
- It jumped the gun by announcing teenage darts player Luke Littler had won the PDC World Championship – before he’d even played in the final.
- In a more serious blunder, it created a fake BBC alert claiming Luigi Mangione, who’s accused of killing UnitedHealthcare CEO Brian Thompson, had killed himself.
- The system stamped The New York Times’ name on a completely made-up story about Israeli Prime Minister Benjamin Netanyahu being arrested.
The BBC, angered over seeing its name attached to fake stories, eventually filed a formal complaint. Press groups joined in, such as Reporters Without Borders, who warned that letting AI rewrite the news puts the public’s right to accurate information at risk.
Die National Union of Journalists also called for the feature to be removed, saying readers shouldn’t have to guess whether what they’re reading is real.
Forschung has previously shown that even when people learn that AI-created media is fake, it still leaves a psychological ‘mark’ that persists afterwards.
Apple Geheimdienst – which offered a range of AI-powered features including AI news – was one of the headline features of the new iPhone 16 range.
Apple is a company that prides itself on polished products that ‘just work’ – it’s rare for Apple to backtrack – so they evidently had little choice here.
That said, they’re not alone as far as AI blunders go. Not long ago, Google’s AI-generated search summaries told people they could eat rocks and put glue on pizza.
Apple plans to resurrect the feature with warning labels and special formatting to show when AI creates the summaries.
Should readers have to decode different fonts and labels just to know if they’re reading real news? And here’s a radical idea – they could just keep displaying the news headline itself?
It all goes to show that, as AI continues to seep into every corner of our digital lives, some things – like receiving accurate news facts – are simply too important to get wrong.
A big U-turn from Apple, but probably not the last we’ll see of its type.
