A Practical Survey on Emerging Threats from AI-driven Voice Attacks: How Vulnerable are Commercial Voice Control Systems?

The emergence of Artificial Intelligence (AI)-driven audio attacks hasrevealed new security vulnerabilities in voice control systems. Whileresearchers have introduced a multitude of attack strategies targeting voicecontrol systems (VCS), the continual advancements of VCS have diminished theimpact of many such attacks. Recognizing this dynamic landscape, our studyendeavors to comprehensively assess the resilience of commercial voice controlsystems against a spectrum of malicious audio attacks. Through extensiveexperimentation, we evaluate six prominent attack techniques across acollection of voice control interfaces and devices. Contrary to prevailingnarratives, our results suggest that commercial voice control systems exhibitenhanced resistance to existing threats. Particularly, our research highlightsthe ineffectiveness of white-box attacks in black-box scenarios. Furthermore,the adversaries encounter substantial obstacles in obtaining precise gradientestimations during query-based interactions with commercial systems, such asApple Siri and Samsung Bixby. Meanwhile, we find that current defensestrategies are not completely immune to advanced attacks. Our findingscontribute valuable insights for enhancing defense mechanisms in VCS. Throughthis survey, we aim to raise awareness within the academic community about thesecurity concerns of VCS and advocate for continued research in this crucialarea.