Apple To Update AI After False Headlines Reported

Apple has said it will update its AI features to clarify when notification summaries are generated automatically, after the summary feature was found to be producing inaccurate information.

The company had faced calls to suspend the summaries, but said it was not planning to do so.

The company’s statement on Monday was the first time it has acknowledged the issue, which produced inaccurate headlines appearing to come from the BBC and The New York Times.

One BBC summary falsely claimed that murder suspect Luigi Mangione had shot himself, while another said Luke Littler had won the PDC World Darts Championship hours before the event took place, and a third said Spanish tennis player Rafael Nadal had come out as gay.

Misinformation

The inaccurate BBC summaries have all appeared since Apple’s AI was released in the UK in December.

In November a ProPublica journalist reported that Apple’s AI produced a false headline claiming Israel prime minister Benjamin Netanyahu had been arrested, when summarising a New York Times headline.

The feature summarises multiple notifications from a single app in order to allow users to “scan for key details”, according to Apple.

The BBC said Apple should “urgently address” the issue.

“These AI summarisations by Apple do not reflect – and in some cases completely contradict – the original BBC content,” the corporation said.

“It is critical that Apple urgently addresses these issues as the accuracy of our news is essential in maintaining trust.”

Reporters Without Borders called on Apple to disable the feature after the headline featuring Luigi Mangione.

‘Further clarity’

The group said the incident showed “generative AI services are still too immature to produce reliable information for the public”.

Apple said it would release an update “in the coming weeks” to “further clarify when the text being displayed is summarisation provided by Apple Intelligence”.

The company said it encourages users to report “unexpected” notification summaries.

The feature, which is available on newer iPhones, iPads and Macs, has been reported on social media to frequently produce overly literal or bizarre summaries for various types of notifications.

Similar issues were reported with Google’s AI search summaries when the company began rolling them out early last year.