Deepfakes Are Becoming Almost Impossible to Detect

Testing the efficacy of AI-powered deepfake detectors, researchers found that they struggled to identify manipulated videos even when multiple layers of editing were applied, as was the case with a recent attempt by a young girl to fool her parents. A 14-year-old girl attempted to create and share fake video calls with her family while they were out of the country. Using AI-powered tools, she created a convincing deepfake that mimicked her voice and face on the screen. However, when her father picked up the phone, he immediately suspected something was amiss. The girl had clearly failed in her attempt to deceive him, but it was not because the technology was flawed – rather, it was simply because her father’s expectations of how a human voice should sound were so deeply ingrained that even a well-made deepfake sounded “like a robot” to him. This experience highlights the increasingly blurred lines between reality and deception in our digital age. As AI-powered tools for creating and detecting deepfakes continue to improve, it is likely that we will see more sophisticated attempts to manipulate audio and video recordings in the future.