Electronic KYCs, known for their convenience and cost-effectiveness, have become integral in various services, replacing the necessity for physical verification. With the widespread availability of high-speed internet and increased accessibility, a multitude of companies is embracing electronic Know Your Customer (e-KYC) methods. The process involves capturing a selfie with an ID or posing in front of the phone's camera, offering swift verification. However, the reliability of these user-friendly procedures is now being tested in the face of advanced AI tools.
E-KYC typically employs two methods: image and video authentication. In the first method, users submit a photo with an ID document, such as a passport or driving license, uniquely linked to them. Human reviewers or algorithms then scrutinize this image against existing records to prevent impersonation. Security features like Machine Readable Zone (MRZ) codes, signatures, stamps, and facial matching enhance authenticity and deter identity spoofing attempts. The second method necessitates individuals to pose with their identity documents before smartphones and computers for live verification.
Recently, concerns have arisen regarding the vulnerability of e-KYC processes to identity spoofing through generative AI tools. Some images posted on a Reddit discussion claimed that fraudsters could exploit generative AI to pass e-KYC with deepfaked identity documents. This involves using open-source AI tools to edit downloaded selfies, creating synthetic images for identity verification, potentially leading to identity theft or fraudulent activities.
Despite efforts to enhance security, concerns persist, especially with the rise of AI-enabled identity spoofing. Liveness tests for ID verification, which require live interactions, are susceptible to deepfake attacks, allowing fraudsters to replace faces with synthetic ones. Even video verification, considered a secure digital verification method, is not foolproof against evolving AI capabilities.
An executive from an Indian verification company emphasized the reliance on human judgment for video-based e-KYCs, introducing a predetermined questionnaire to check the liveliness of individuals. However, the quality of deepfakes used in verification significantly impacts the reviewer's decision, posing challenges when dealing with old or poor-quality images in identity documents.
The prevalence of e-KYC has surged, driven by the convenience it offers, particularly during the Covid-19 pandemic. The global e-KYC market, valued at USD 625.06 million in 2022, is anticipated to reach USD 2.23 billion by 2028. While Aadhaar-enabled e-KYC transactions exceeded 14.95 billion by April 2023, concerns about AI's ability to create realistic images and the potential for identity fraud using deepfake technology loom large.
Despite the cost reduction in customer onboarding facilitated by e-KYC, the shadow cast by AI's capabilities and insufficient safeguards in generative models poses a growing threat to the e-KYC market. Identity fraud using deepfake technology has recorded a significant rise, impacting countries like India, which is among the top ten in the Asia-Pacific region most affected by such cybercrimes. A report from an Indian online verification company revealed mismatches and liveness check failures in a considerable percentage of loan applicants.
