OpenAI CEO Sam Altman has warned financial institutions of fraud risks enabled by AI, urging reform in authentication and wider protections against systemic threats.
The warning was delivered at a recent event hosted by the Federal Reserve in Washington, DC, where he voiced concerns over the industry's current preparedness.
A key issue raised was the continued reliance by some financial institutions on voice-based authentication for high-value transactions. Sam stated that AI tools are now capable of convincingly replicating human voices, rendering such methods vulnerable to manipulation. He described the practice as outdated and inadequate given current AI capabilities.
Concerns over large-scale attacks and evolving fraud tactics
Altman also pointed to the broader implications of generative AI in fraud. As synthetic content becomes more realistic, including deepfake video calls, existing verification systems may no longer be sufficient. He suggested that banks and regulators should rethink how identity and intent are verified, proposing that both industry and the public need to be educated on emerging fraud tactics and safer methods of communication.
The potential for foreign entities to deploy AI tools in coordinated attacks on financial infrastructure was also raised as a growing concern. Altman expressed that the industry should prepare for scenarios in which large-scale fraud or system breaches could be orchestrated using advanced AI techniques.
These comments come in the context of increasing integration of AI into the financial sector, with firms adopting the technology to enhance customer engagement and operational efficiency.
The remarks also reflect a wider debate in the AI research community. Figures from both OpenAI and Anthropic have recently criticised competing firms, particularly xAI, for what they describe as lax safety standards. These criticisms could influence ongoing discussions about regulatory oversight and enterprise use of AI tools.
Researchers and technology leaders, including those involved in the early development of AI systems, have repeatedly called for greater scrutiny and caution around the deployment of advanced models. The warnings highlight a growing consensus that while AI offers substantial opportunities, its risks, especially in finance, should not be underestimated.