As cited
Copy frozen at (site build).
ai security
Risky Bulletin: Voice cloning defenses still weak, can be bypassed
Researchers from the University of Texas at San Antonio demonstrated that current voice cloning defenses, which rely on injecting noise into audio recordings, can be bypassed by attackers who account for the added noise. While voice cloning attacks currently produce detectable low-quality output, the researchers argue that existing defense mechanisms are insufficiently complex and vulnerable to more sophisticated approaches.
Why it matters: Organizations relying on voice biometrics and voice-based authentication systems should reassess their defenses against cloning attacks, as current noise-injection protections may not provide adequate security if attackers adapt their techniques.
- Source published
- First seen by Cybersecurity Tracker