'CBS Mornings' hosts Gayle King and Nate Burleson attempt to tell real voices from deepfakes: 'It's really scary'
NEW YORK CITY, NEW YORK: Artificial Intelligence has been having eureka moments every now and then, and while it is exciting for many, it remains a concern for others. 'CBS Mornings' made a report on the recent update in the world of AI and even the hosts were left quite worried.
In the episode aired on the morning of August 3, 2023, Natalie Morales brought up the popular AI scam that has been going around the world using the tactic of deepfake voices that have developed to sound really close to real voices. Co-hosts Nate Burleson and Gayle King became Morales' test subjects in finding out how similar AI-generated voices are to real ones.
What is the deepfake voice scam?
In a recent case that shocked not just the country but the whole world, a woman was almost severely scammed for thousands of dollars. Jennifer DeStefano received a call from what sounded like her teenage daughter Brianna. "Brianna's" voice sounded like she was in a panicked state and she asked her mother to help and save her from "these bad men." Following that, some men got on the call, confirming that they had DeStefano's daughter and demanded a million dollars for ransom. The call was discovered to be a scam and there was no case of kidnapping behind the call, just AI-generated voices of the daughter and the scammers.
Nate Burleson and Gayle King's skills tested
Fill-in host for 'CBS Mornings', Morales, brought up the disturbing study about deepfake speech in a segment for the breakfast show. She reported that the study found that people were not able to tell real and AI-generated voices apart one out of four times, which is a major change from the previous quality of these fake voices.
Can @GayleKing and @NateBurleson tell the difference between a real voice and a deepfake?
— CBS Mornings (@CBSMornings) August 3, 2023
A disturbing new study found that one in four people were fooled by deepfake speech. https://t.co/vmVY912EZw pic.twitter.com/ir6cLA5gPy
Morales asked Burleson and King to tell whether the two samples of voices they had were fake or real. While both the recordings sounded fairly real, the two hosts weren't able to correctly tell them apart, one of them being AI-generated. "They are capturing emotions now too which is really scary," Morales pointed out as the hosts noticed how the voices were able to sound human because of the subtle inflections. All three hosts agreed that it was quite scary.