66-year-old grandma shocked as AI sends message filled with profanity: ‘Been able to have sex’
A Scottish woman was left stunned after Apple’s AI-powered voice-to-text software mistakenly inserted offensive language and explicit content.
A shocking error in Apple’s AI dictation software left an elderly Scottish woman appalled after it inaccurately transcribed a voicemail with offensive language and inappropriate sexual references, reported the New York Post.
Louise Littlejohn, 66, was stunned when she received the message from Lookers Land Rover garage in Motherwell, inviting her to an event. However, the iPhone’s voice-to-text service completely misinterpreted the call, referring to her as a “piece of s–t” and asking if she had “been able to have sex.”
“The text was obviously quite inappropriate,” Littlejohn told the post, recalling the shocking error.
Initially, she assumed the message was a scam, but after recognizing the zip code, she realized it was from the car dealership where she had previously made a purchase.
“The garage is trying to sell cars, and instead of that they are leaving insulting messages without even being aware of it,” she explained. “It is not their fault at all.”
Experts suggest the error may have been caused by the caller’s Scottish accent, but background noise and the scripted nature of the message were likely bigger factors in the mix-up.
Peter Bell, a professor of speech technology at the University of Edinburgh, noted that multiple elements contribute to transcription errors.
“All of those factors contribute to the system doing badly,” Bell told the Daily Mail.
Misinterpretation
Some analysts believe the word “sex” may have been a misinterpretation of “sixth,” referring to the event’s date—March 6.
Despite the bizarre blunder, Littlejohn has taken it in stride.
“Initially, I was shocked — astonished — but then I thought that is so funny,” she said.
However, the incident raises concerns about the reliability of AI-driven transcription services.
“The bigger question is why it outputs that kind of content,” Bell pointed out. “If you are producing a speech-to-text system that is being used by the public, you would think you would have safeguards for that kind of thing.”
Apple has faced similar issues before. Just last month, the company drew backlash from MAGA supporters when its software mistakenly transcribed the word “racist-” as “Trump.” Apple later clarified that its dictation feature sometimes displays words with phonetic similarities before self-correcting.
