OpenAI CEO Sam Altman warned during his speech at the New York Fintech Conference on July 22, 2025, that the financial sector is battling an unprecedented fraud crisis as advanced AI technologies can accurately mimic customers' voices, bypassing voice-based biometric security systems. According to Altman, financial institutions reported $7.8 billion in losses from fraud committed using voice-mimicking AI technologies in just the first half of 2025, representing a 340% increase compared to the same period last year. The problem is particularly severe as 78% of financial service providers use some form of voice-based authentication system, and 72% of current technologies cannot reliably detect AI-generated voices, as demonstrated by a study conducted by Carnegie Mellon University in April 2025.
Criminal groups have been exploiting AI voice-mimicking technology with increasingly sophisticated methods since the second half of 2023, and today a $400 voice-cloning package is available that allows anyone to almost perfectly reproduce a person's voice based on just a 3-minute voice sample. According to Federal Trade Commission data, 64% of victims of such scams are older clients over 65 years of age, and the average fraud amount reaches $17,300 per person. Sarah Johnson, Security Director of the Financial Services Association, stated that voice-based security systems alone are no longer sufficient, and 83% of banks are already working on implementing multi-factor security solutions that include machine learning models to distinguish between real and artificial voices.
Altman urged financial institutions to implement advanced, AI-based deep voice analysis systems by the end of 2026, capable of identifying synthetic voices with 94.7% accuracy, as well as employing multi-layered security protocols including behavioural biometrics and transaction anomaly detection. OpenAI itself is actively working on a solution and announced the launch of the VoiceGuard system in September 2025, which will be freely available to financial institutions and can identify voices generated by the company's own models. Meanwhile, regulatory authorities are also taking steps: the Securities and Exchange Commission (SEC) is expected to issue new guidelines in August 2025 requiring financial institutions to improve their security protocols and prepare detailed reports on AI fraud attempts.
Sources:
The Associated Press
Clare Duffy
OpenAI’s Sam Altman has warned that voice authentication is increasingly vulnerable to AI-generated impersonation, urging financial institutions to urgently upgrade their security systems to prevent a looming fraud crisis.
