Deepfakes and the AI Arms Race in Bank Cybersecurity

Fed Governor Michael S. Barr delivered a speech on April 17th at the Federal Reserve Bank of New York entitled "Deepfakes and the AI Arms Race in Bank Cybersecurity."

Governor Barr discussed the large and growing threat of GenAI-driven cybercrime, specifically focusing on deepfakes created using generative adversarial networks (GANs). These deepfakes produce highly convincing synthetic voice and video impersonations, targeting high-trust channels like executive communications and customer service.

To combat this, he suggested we utilize enhanced identity verification (biometrics, voice analytics), AI-driven transaction monitoring, stronger customer authentication, and AI tools for detecting and defending against synthetic media.

Barr spoke of the attack asymmetry. Attackers can develop and deploy new deepfake techniques much faster than defenders can adapt, leading to a potential imbalance.

These deepfakes produce highly convincing synthetic voice and video impersonations, targeting high-trust channels like executive communications and customer service.

The systemic costs are high, he said. Even a single successful deepfake attack can have far-reaching consequences, potentially damaging reputations, compromising sensitive data, and eroding trust. The speech is HERE