This paper presents a comprehensive review of the emerging threat landscape targeting voice authentication systems (VAS) and anti-spoofing countermeasures (CMs). It traces the significant shift in voice authentication from traditional systems relying on handcrafted acoustic features to deep learning models capable of extracting robust speaker embeddings. It addresses a variety of attack types, including data poisoning, adversarial attacks, deepfakes, and adversarial spoofing attacks. For each attack type, it summarizes methodologies, highlights commonly used datasets, compares performance and limitations, and organizes the existing literature using a widely accepted taxonomy. By highlighting emerging risks and unresolved challenges, it aims to support the development of more secure and resilient voice authentication systems.
