A voice replay attack captures genuine speech from the target and plays it back to defeat voice authentication or liveness checks. No AI generation is required; the attacker exploits trust in audio without needing to fake it.

• Real audio is recorded covertly or pulled from public sources
• The recording is played back through a speaker or injected into a call to satisfy an authentication prompt
• Some systems can be fooled even when the prompt requires specific phrases, if attackers stitch together captured fragments

• Voice biometric authentication bypass at banks and contact centers
• Liveness check failure during remote KYC and onboarding
• Compounding attacks where replayed audio is processed through speakers and re-recorded to imitate real-world acoustics, masking synthetic artifacts that detectors usually flag
• Pretext calls that use captured voicemail snippets to impersonate the target