Apple is struggling with issues related to AI Summaries, part of Apple Intelligence, which aim to smartly summarize notifications on iPhones, iPads, or Macs. While it often works, sometimes the system, still in beta like Apple Intelligence itself, makes errors that nearly resemble fake news. Recently, the British broadcaster BBC complained to Apple about this. Journalist organizations even called for the service to be completely shut down, claiming it poses a “threat to journalism.” Apple has now addressed the issue and promised a partial fix via an update, but the solution is mild.
In a statement to the BBC, Apple announced a “software update in the coming weeks” to ensure that text generated by AI Summaries will be “offered by Apple Intelligence.” Currently, this is surprisingly not the case. The AI-generated summaries are only marked with a simple icon consisting of an arrow and two text lines. Apple encourages users to report their concerns if a notification summary seems “unexpected.”
Apple’s solution highlights a problem also faced by competitors like OpenAI, Google, or Anthropic. They regularly mention in fine print that the system’s output should be “verified.” However, few users actually do this, and many rely on or consider the AI output “good enough,” even when it is incorrect. Technically, it is only partially possible to eliminate erroneous outputs from the systems, as the hallucination is technically inherent.
Apple also informed the BBC that it is “continuously making improvements using user feedback.” Additionally, receiving AI Summaries is “optional.” The BBC had complained that Apple Intelligence turned a headline into a death notice for an attacker who is still alive. Other misinterpreted notifications included a false dart world champion and a tennis player who allegedly came out. “These AI summaries from Apple do not reflect the original BBC content or are completely wrong in some cases,” the BBC stated.
It is now critically important for Apple to address these issues because “the accuracy of our news is essential for people to have trust.” The BBC is not the only “victim.” Other media outlets like the New York Times were also affected. Even simple summaries of chat conversations can go awry. For instance, a challenging hiking tour was summarized as an alleged “suicide.” On Reddit, there is now a whole subreddit dedicated to Apple Intelligence fails.