Untapped AI Voice Authentication Patent: USPTO Filing Targets Fraud

• South African company Untapped AI has filed two patent applications with the USPTO for real-time voice authentication technology.
• The technology aims to detect and terminate calls where the voice is identified as inauthentic, combating AI voice fraud.
• South Africa experienced R2.4 billion in digital banking crime in 2025, largely due to social engineering, before widespread AI voice fraud.
• High-profile deepfake incidents include a €95 million fraud against Fideuram and a US$25 million loss for Arup, both involving AI-cloned voices.
• The patent addresses a critical need for enhanced security as voice cloning tools become more accessible to organized crime.

The Untapped AI USPTO filing represents a crucial development in the battle against sophisticated digital fraud.

A South African artificial intelligence firm, Untapped AI, has taken a significant step to counter the escalating threat of voice-based fraud. In September, the company submitted two patent applications to the United States Patent and Trademark Office (USPTO). These filings introduce an Untapped AI voice authentication patent designed to verify the identity of a caller in real-time during an active phone conversation.

The core functionality of this innovative deepfake fraud prevention technology involves authenticating the voice on the line and automatically terminating the call if the voice is determined to be fraudulent. This development directly addresses a critical vulnerability in modern communication: the inability to reliably trust a voice heard over the phone, a problem exacerbated by the widespread availability of sophisticated voice cloning tools. The technology aims to provide a robust defense for sectors such as banking, insurance, telecommunications, and government, where secure voice authentication is paramount.

The urgency for such advanced security measures is underscored by the alarming increase in digital banking crime. According to the South African Banking Risk Information Centre's (SABRIC) 2025 Annual Crime Statistics Report, South Africans collectively lost R2.4 billion to digital banking crime in 2025, across more than 110,000 reported incidents. A substantial portion of these offenses were driven by social engineering and the manipulation of customers, rather than direct breaches of bank systems, indicating a significant human element in the fraud landscape.

These considerable losses occurred prior to AI voice fraud South Africa reaching its full potential. With voice cloning tools now freely accessible and capable of generating thousands of simultaneous scam calls, often disguised by spoofed numbers that mimic legitimate institutions, Untapped AI anticipates a sharp rise in fraudulent calling and voice impersonation as these tools increasingly fall into the hands of organized crime. The nation's current defenses are largely insufficient; many South African call centers lack the capability to detect cloned voices, finance teams struggle to verify the authenticity of urgent payment approvals from senior executives, and customers often cannot confirm if a call from their bank is genuine, leaving wide gaps for criminals to exploit.

The devastating financial and reputational consequences of banking voice cloning security breaches are evident in recent global incidents. In February of the current year, fraudsters successfully siphoned €95 million, approximately R1.9 billion, from Fideuram, the private banking division of Italy's largest lender, Intesa Sanpaolo. This elaborate scheme, reported by Reuters, commenced with a WhatsApp message seemingly from the group's chief executive and culminated in a phone call where criminals utilized AI to clone the voice of a senior partner at a prominent law firm. The bank's then-chairman, convinced by the cloned voice, authorized a series of transfers to accounts primarily located in China and Hong Kong, with €36 million remaining unrecovered after being moved through international accounts and converted into cryptocurrency.

Despite Fideuram benefiting from the extensive resources and security teams of one of Europe's largest banking groups, a single cloned voice bypassed their defenses. Other notable incidents include a 2024 case where an employee at global engineering firm Arup's Hong Kong office participated in a video call with AI-generated fakes of the company's CFO and colleagues, leading to 15 transfers totaling US$25 million. In 2025, an impostor used an AI clone of US Secretary of State Marco Rubio's voice to leave messages for foreign ministers and American officials, while in the same year, fraudsters imitated an Italian government minister to persuade businessman Massimo Moratti to transfer nearly €1 million to an overseas account. The impact extends beyond corporate deepfake liability, affecting individuals directly, as demonstrated by Arizona mother Jennifer DeStefano, who received a ransom call featuring an AI-cloned voice of her 15-year-old daughter, who was, in fact, safe on a ski trip. DeStefano later testified about this harrowing experience before the US Senate.

The Untapped AI USPTO filing represents a crucial development in the battle against sophisticated digital fraud. As AI voice cloning technology becomes more accessible and potent, the risks of financial loss, reputational damage, and erosion of trust for both institutions and individuals are escalating rapidly. The patent application for real-time voice authentication offers a potential technological safeguard against these evolving threats, particularly in high-value transactions and critical communications where voice verification is paramount.

This highlights the urgent need for legal and compliance teams to reassess and update existing fraud prevention protocols, internal controls, and employee training programs. The increasing prevalence of deepfake attacks, as evidenced by the Fideuram and Arup cases, underscores the significant corporate deepfake liability that companies face if they fail to implement robust deepfake fraud prevention technology. Proactive advisory for clients on these emerging risks and the available technological and procedural countermeasures is becoming indispensable in navigating this complex and rapidly changing landscape.

Legal and compliance teams must urgently review and update fraud prevention protocols, internal controls, and employee training to mitigate significant financial and reputational risks posed by sophisticated AI voice cloning and deepfake attacks, particularly in financial transactions and corporate communications. The emergence of patented solutions like Untapped AI's indicates a developing legal and technological landscape for fraud liability and prevention, requiring proactive advisory for clients.

Source: Original reporting via industry news.