Low-Tech Solutions Recommended to Combat Advanced AI Deepfakes in Corporate Fraud

Deepfakes Evolve Beyond Detection

AI-enabled identity theft has become highly sophisticated, making it difficult to identify deepfakes through traditional tells. Advanced attacks can involve long-term infiltration of company systems, as demonstrated by a January 2024 incident where an Arup employee participated in a video call with AI-generated clones of executives, leading to $25 million in wire transfers.

Deepak Gupta, technical CEO at GrackerAI, stated that recognizing a face and voice is no longer sufficient proof of identity, as deepfake technology has broken these protocols. The technology has outpaced previous detection methods, such as background noise or lack of breathing sounds, making it challenging to discern real individuals from AI fakes.

The Erosion of Traditional Verification

Live voice and video calls have historically been the standard for identity verification in corporate settings, especially for high-value transactions or sensitive data exchanges. However, deepfake technology has eroded this standard over the past five years, with AI models rapidly improving their ability to mimic human behavior.

Low-Tech Protocols as a Solution

Security experts, including Gupta and James Scobey, CTO at B2B cybersecurity firm S2i2, suggest that the solution lies in adopting low-tech security protocols. These methods, once considered too simple for large-scale business operations, are now seen as more effective defenses against advanced AI deepfakes.

Impact on Businesses

The rise of undetectable deepfakes poses a significant financial risk to companies, as evidenced by the Arup incident. Beyond direct financial losses, the damage to customer trust and potential fines for legislative non-compliance can severely impact a company's profitability and long-term viability.