What are deepfakes?
Deepfakes are videos manipulated by AI to put words in people’s mouths and distort the truth. The tools used to create deepfakes are becoming increasingly accessible to the general public, as well as becoming more sophisticated.
Deepfakes are often created for satirical or darkly comedic purposes. Earlier this year, a deepfake video showed Mark Zuckerberg gleefully boasting about his ownership of millions of people’s stolen data. However, as deepfake technology advances, there is a concern that technology is being used maliciously.
David Wright, the director of Trilateral Research believes that people should be deeply concerned about the issue. He believes as technology evolves it will become increasingly difficult to distinguish what is real from what is not. Deepfakes could, therefore, be used in several malicious ways, such as ultra-sophisticated spear phishing, spreading fake news, and deceiving the general public.
In an attempt to get ahead of the increasingly sophisticated deepfake technology, researchers have been exploring alternative methods. A team at the University of Oregon’s Institute of Neuroscience is working on training mice to understand irregularities within speech, a task that animals can do with remarkable accuracy.
The mice have been trained to categorize different speech sounds and, therefore, it should be possible to train the mice to detect real and fake speech.
The mice could correctly identify speech sounds approximately 80% of the time. While this is not perfect, when coupled with existing methods of detecting deepfakes, it could be an extremely valuable input.
The goal of the experiment will be to take the lessons that are learnt from the way the mice detect fake speech and implement it in a computer. The systems will then be used to help identify deepfake videos.
Have you been tricked by a deepfake video? Let us know on Twitter @Hyve!