Congress and several states have responded with a suite of bills and laws aimed at containing AI-driven fraud, which generated an estimated $12.5 billion in losses in 2024. The legislative reaction seeks to slow economic damage and restore institutional trust through tougher penalties, crypto-focused measures and whistleblower protections.
Lawmakers push criminal, operational and governance tools to curb AI fraud growth
The federal package outlines proposals that significantly raise sanctions and expand enforcement capabilities. The AI Fraud Deterrence Act (Nov. 25, 2025) proposes doubling fines up to $2 million and introducing prison terms of up to 30 years when AI is involved in wire fraud, bank fraud or money laundering.
Another initiative, the Preventing Deep Fake Scams Act (June 27, 2025), would establish a task force to guide financial institutions in adopting defensive AI and consumer-protection protocols. This dual strategy blends criminal deterrence with operational resilience across the financial system.
Lawmakers have also targeted risk channels linked to digital assets. The Crypto ATM Fraud Prevention Act proposes transaction limits for new users — $2,000 daily and $10,000 over 14 days — along with mandatory fraud notices and refund mechanisms tied to police reports, responding to $114 million in losses from ATM-related scams in 2023.
In parallel, the TAKE IT DOWN Act (May 19, 2025) introduces penalties for distributing nonconsensual intimate images, including those produced with generative AI. The law expands legal protection against reputational and personal harm driven by synthetic media.
There is also an institutional governance layer: The AI Whistleblower Protection Act (June 2, 2025) safeguards employees who report violations or risk linked to AI systems and challenges contractual barriers that suppress disclosure.
States have advanced outside the federal framework as well. Tennessee passed the ELVIS Act, extending rights of publicity to voice and image with penalties of up to 11 months and 29 days in jail and $2,500 fines for unauthorized voice or image cloning. Utah, meanwhile, expanded in 2024 the definition of child sexual abuse material to include AI-generated imagery.
Texas continues to debate frameworks addressing behavioral manipulation, impact assessments and design restrictions. For financial institutions, effective fraud detection must pair AI tools with traceability, digital forensics and custody trails to support prosecution.
Experts warn that technological pace compounds the challenge. Hany Farid’s remark that “AI years are dog years” highlights the speed of evolution, while Mohith Agadi stresses that without forensic investment, proving AI involvement remains difficult.
The regulatory response has broadened across criminal, operational and institutional dimensions. Future effectiveness depends on integrating legal mandates with evidence systems capable of tracing, detecting and attributing AI-enabled fraud.
