acm-header
Sign In

Communications of the ACM

ACM News

Scientists Jack Into Subjects' Brains, Find That They Can Identify Deepfakes Subconsciously


Although more research is needed, the findings presage the intriguing possibility of tech that leverages the immense processing power of our own brains—by identifying knowledge of which we're not even consciously aware.

Credit: Getty Images/Futurism

As deepfakes — AI powered likenesses of human faces and voices — grow increasingly prevalent, the ability to detect them becomes more imperative.

But a provocative new study claims that our brains can subconsciously detect deepfakes, even if our conscious minds are deceived.

Conducted by a team of researchers from the University of Sydney, the findings indicate that subjects' brains, which were being monitored using electroencephalography (EEG), were successfully able to detect deepfakes 54%  of the time. But when an earlier group was asked to verbally identify the same deepfakes, their success rate was only 37%.

From Futurism
View Full Article

 


 

No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account