Advances in generative artificial intelligence (AI) have enabled authentic-sounding speech synthesis (语音合成) to the point that a person can no longer distinguish whether they are talking to another human or a deepfake (深度伪造). If a person's own voice is "cloned" by a third party without their agreement, bad guys can use it to send any message they want.
Computer scientist and engineer Ning Zhang has developed a new method to prevent unauthorized speech synthesis before it takes place: a tool called AntiFake.
Traditional deepfake detection methods only work after the damage is done. However, AntiFake prevents voice data from being synthesized into an audio deepfake beforehand. This tool turns the tables on cybercriminals (网络罪犯) by using similar voice cloning techniques they employ, but for voice protection. The software complicates voice data extraction and feature identification crucial for voice synthesis. "We're using an opposite AI technique originally used by cybercriminals, but now we're using it against them," Zhang explained. "We slightly change the recorded audio, just enough to make it unusable for voice clone training while still sounding natural to humans."
Ben Zhao, a professor of computer science at University of Chicago, says that the software, like all digital security systems, will never provide complete protection. But, he adds that it can raise the bar and limit the attack to a smaller group of individuals with significant resources.
AntiFake can already protect shorter voice recordings against cloning. The creators of the tool believe that it could be extended to protect larger audio documents or music from misuse. But the methods and tools that are developed must be continuously adapted because cybercriminals will learn and grow with them.