Deepfake attacks are becoming a significant threat to contact centers, according to a report from Pindrop, a company focused on voice technology that works with companies in the financial services industry. The report highlights a worrying trend: fraudsters are increasingly using deepfake technology to deceive contact center agents and commit financial fraud.
Deepfake technology, which uses AI to create convincing fake audio and video, has become more accessible and effective. This technology poses a severe risk to financial institutions, where fraudsters can impersonate clients convincingly enough to conduct unauthorized transactions. Pindrop’s report reveals that 67% of respondents in the banking and finance sector are concerned about deepfake and voice cloning attacks.
Contact centers, especially those handling high-net-worth individuals, are prime targets. Fraudsters use deepfake voices to bypass traditional security measures, tricking agents into performing actions such as wire transfers, ordering new credit cards, or changing account information. These sophisticated attacks can result in significant financial losses, as seen in a recent case where one company lost $25 million to a deepfake scam. The attackers can still strike even if they have less information about a banking customer. In one instance, fraudsters used a synthetic voice with an accent to pass as a customer who is a Latin American woman.
To protect against these sophisticated attacks, contact centers need to adopt several proactive measures:
- Liveness Detection Technology: Implementing advanced liveness detection can help differentiate between real and synthetic voices. This technology analyzes audio patterns to identify signs of deepfakes, such as uniform pauses or spectral distortions, which are hard for machines to replicate consistently.
- Integrated Multi-Factor Authentication (MFA): Combining voice authentication with other factors like device recognition, behavioral analysis, and carrier metadata increases the difficulty for fraudsters to deceive the system. This multi-layered approach enhances the accuracy of fraud detection and reduces the risk of successful deepfake attacks.
- Continuous Fraud Detection: Implementing systems that continuously monitor for signs of fraud can help identify and mitigate threats early. This includes using voice mismatch detection to flag when a caller’s voice does not match the enrolled voice for an account, which can indicate potential fraud.
- Early Risk Detection: Starting the risk assessment process at the beginning of a call lifecycle is crucial. Techniques such as caller ID spoof detection and DTMF keypress analysis can provide early indicators of fraudulent activity, allowing contact centers to take preventive action promptly.
- Employee Training and Awareness: Regular training programs for contact center agents on the latest fraud tactics and how to recognize deepfake attacks are essential. Agents should be equipped with the knowledge to identify suspicious behavior and escalate potential fraud cases appropriately.
