AI Fraud Is Slipping Through a Gap in US Export Controls, Experts Warn

_{Photo credit: Solen Feyissa / Unsplash.com}

US efforts to control the global spread of artificial intelligence have focused heavily on advanced chips and frontier AI models. But according to a new analysis highlighted by Cybernews, another category of AI may be receiving far less scrutiny: tools that can be used for fraud, impersonation, and social engineering.

The report argues that current export controls are designed primarily around national security concerns, particularly restricting access to the computing power needed to train advanced AI systems. While those measures target strategic competitors, they do little to address the international availability of AI-powered fraud tools.

Currently, many of the techniques increasingly used in online scams do not require cutting-edge infrastructure, thanks to tools such as voice-cloning software, phishing content generators, synthetic identity tools, and automated social engineering systems. These tools are widely available and relatively inexpensive to deploy.

Regulators continue tightening restrictions on AI-related exports. In recent years, Washington has introduced multiple measures that limit access to advanced semiconductors and high-performance computing technologies, particularly in relation to China.

However, the report states that the same policy discussions rarely focus on how AI is being used in financial crime. Fraudsters have already adopted generative AI to create fake documents, imitate voices, and automate scam campaigns, all of which previously required significant manual effort.

According to the analysis, there’s a growing disconnect between the types of AI attracting regulatory attention and the ones that are already used in everyday criminal activity. Even as governments continue to develop AI policies, the report argues that fraud risks may require their greater attention alongside the geopolitical and national security concerns.