A new report has revealed a staggering 3000% increase in deepfake-based identity fraud attempts in 2023 compared to the previous year. The findings, detailed in Onfido’s 2024 Identity Fraud Report, highlight a significant shift in criminal tactics, with artificial intelligence supercharging the creation of synthetic media for fraudulent purposes. This surge in AI-driven attacks occurred while rates of traditional identity document and selfie fraud remained stagnant.

AI Supercharges Forgery Attempts

The report indicates that generative AI has lowered the barrier to entry for criminals, enabling even those with limited technical skills to create convincing fake documents and deepfakes. This resulted in deepfakes accounting for 3% of all fraud attacks in 2023, a sharp rise from just 0.1% in 2022. The attacks often involve ‘face swaps’ or presenting a screen displaying a video of a victim during live identity verification checks. Furthermore, the detection rate for ‘easy’ forgeries, defined as digitally altered documents, saw a five-fold increase during the same period. According to Onfido fraud specialist Simon Horswell, AI has “supercharged” the ability for criminals to commit fraud at scale.

Global Trends in Document and Selfie Fraud

While AI-powered fraud skyrocketed, traditional identity fraud rates plateaued, with document and selfie fraud remaining consistent year-over-year at approximately 5% of all identity checks. Geographically, the report found that the United Kingdom had the highest rate of document fraud at 6.8%, while the United States experienced the highest rate of selfie fraud at 2.2%. Passports were the most frequently attacked form of identification, yet they also had the lowest fraud rate (3.4%), suggesting a higher level of security. In contrast, national ID cards had the highest fraud rate at 6.1%. The report underscores the role of biometric verification technology in detecting these sophisticated attacks, including its ability to identify screen replays and masks.

Source: https://www.infosecurity-magazine.com/news/ai-deepfake-fraud-skyrockets/