Apple Faces Pressure to Reassess ‘Runaway’ AI News Notifications

Apple is encountering renewed demands to eliminate its contentious artificial intelligence (AI) functionality that has produced erroneous news alerts on its latest iPhones.

This feature aims to condense breaking news notifications but has, in some cases, generated entirely false narratives.

The BBC initially raised objections to the technology giant regarding the misrepresentation of its journalism in December; however, Apple did not address the issue until Monday this week, asserting that it was striving to make clear that the summaries were produced by AI.

Alan Rusbridger, the previous editor of the Guardian, informed the BBC that Apple needed to take more significant action and remove a product he described as “clearly not ready.”

Rusbridger, who is also a member of Meta’s Oversight Board that evaluates appeals concerning the company’s content moderation decisions, remarked that the technology was “out of control” and represented a substantial risk of misinformation.

“Confidence in news is already quite low without massive American corporations entering the fray and using it as a sort of test product,” he stated on the Today programme, aired on BBC Radio Four.

The National Union of Journalists (NUJ), regarded as one of the largest unions for reporters globally, stated that Apple “must act promptly” and eliminate Apple Intelligence to prevent misleading the public – mirroring previous pleas by the journalism organization Reporters Without Borders (RSF).

“At a moment when access to accurate reporting is crucial, the public should not be put in a position of having to second-guess the veracity of the news they receive,” remarked Laura Davison, general secretary of NUJ.

RSF further stated that Apple’s response was inadequate and reiterated its demand for the feature to be taken offline.

The BBC lodged a complaint last month after an AI-generated summary of its headline inaccurately indicated that Luigi Mangione, the individual purportedly responsible for the death of UnitedHealthcare CEO Brian Thompson, had taken his own life.

On Friday, Apple’s AI incorrectly summarized BBC app notifications claiming that Luke Littler had triumphed in the PDC World Darts Championship several hours before it commenced – and that the Spanish tennis star Rafael Nadal had publicly come out as gay.

This instance marks the initial formal response from Apple regarding the concerns expressed by the BBC about the inaccuracies, which seem to originate from within the organization’s app.

“These AI-generated summaries by Apple do not align with – and in some cases completely contradict – the original BBC material,” the BBC declared on Monday.

“It is imperative that Apple urgently rectifies these issues, as the accuracy of our news is vital for sustaining trust.”

The BBC is not the sole news entity affected.

In November, a ProPublica journalist brought attention to erroneous Apple AI summaries of alerts from the New York Times app suggesting it had reported the arrest of Israel’s Prime Minister Benjamin Netanyahu.

Additionally, an inaccurate summary concerning a New York Times article appears to have surfaced on January 6, associated with the fourth anniversary of the Capitol riots.

The New York Times has opted not to comment.

RSF indicated that the misleading, AI-generated headline regarding Mr. Mangione back in December illustrated that “generative AI services remain too immature to provide dependable information for the public.”

On Tuesday, it noted that Apple’s strategy to enhance the feature to clarify when notifications are summarized by AI to users “doesn’t resolve the issue whatsoever”.

“It merely shifts the responsibility onto users, who – in an already perplexing information terrain – will be expected to verify the truthfulness of the information,” stated Vincent Berthier, head of RSF’s technology and journalism division.

Apple announced that its update would be available “in the upcoming weeks”.

It has previously stated that its notification summaries – which consolidate and rephrase previews of several recent app notifications into a single alert on users’ lock screens – are intended to help users “scan for crucial details”.

“Apple Intelligence features are in beta, and we are constantly enhancing them with input from user feedback,” the company articulated in a statement on Monday, noting that receiving the summaries is voluntary.

“A software update in the coming weeks will further clarify when the information shown is a summary provided by Apple Intelligence. We urge users to report any concerns if they encounter an unexpected notification summary.”

The feature, along with other tools released as part of its wider suite of AI resources was deployed in the UK in December. It is solely accessible on its iPhone 16 models, iPhone 15 Pro and Pro Max devices operating on iOS 18.1 and above, as well as on select iPads and Macs.

Apple is not the only entity that has launched generative AI tools capable of creating text, images, and additional content upon user request – although with different outcomes.

Google’s AI summaries functionality, which delivers a textual summary of information from the results at the forefront of its search engine in reaction to user inquiries, encountered criticism last year for generating some inconsistent results.

At that time, a Google spokesperson stated that these were “isolated instances” and that the feature was, in general, functioning properly.