Translated by Authme
With the development of apps and the sharing economy, compounded by the impact of the pandemic, verifying the identity of remote connections without seeing the person has become a significant challenge for businesses. Additionally, the protection of personal information and privacy in digital services is increasingly emphasized. In recent years, the frequent occurrence of identity fraud using Deepfake technology has emerged as a critical challenge that must be addressed in digital identity verification.
Founded in 2019, Authme was established to address this very issue. The founding team, with backgrounds as white-hat hackers and long-term involvement in cybersecurity and technology issues, aims to create convenient, efficient, and secure identity verification solutions through technology, enabling everyone to equally enjoy and experience digital life. They have developed their own eKYC solution, which combines document verification technology, Mobile NFC chip reading, facial recognition, and live detection engines. This system can complete the identity verification process within 3 minutes. By analyzing facial skin texture and blood flow, it helps enterprises effectively defend against identity fraud and counterfeit attacks such as Deepfake.
Starting from the User Experience, Authme Optimizes Its Product Technology
Andy Lee, CEO of Authme, states that in response to Deepfake synthetic images, Authme has integrated live detection technology with an AI engine to recognize altered facial images. He explains that Deepfake impersonation attacks, which are forged through video or selfie images, can occur in two scenarios. One is where the video transmission to the server is intercepted and altered to a fake image, and the other is when the image in front of the video camera is already forged. The approaches to handle these are different. The former requires using a Deepfake detection model when the server receives the image, while the latter needs live detection technology to examine the image in front of the camera. Many details, including the type of camera technology used, such as whether it includes infrared technology or just RGB imaging, need to be discussed. The authenticity is determined through factors like facial blood flow, skin texture, image reflectivity, or other characteristics.
Apart from facial recognition, document verification is also a common method of identity verification. While preventing fraud, matching the aged photos on the documents with the actual person is a challenge that must be addressed. Andy explains that after puberty, the skeletal structure of the face becomes more stable, with changes mainly in height or body shape. By using OCR technology to automatically extract information from documents and comparing it with the face, even photos from thirty years ago in junior high can be recognized. Besides age differences, many real-world considerations, like damaged documents or glare during photo-taking, are common and must be technically overcome.
To enhance immediacy and optimize user experience, reducing user impatience during account opening, insurance applications, or other identity verification processes, Andy mentions that the team also uses Edge AI technology to deploy lightweight models on multiple platforms, allowing users to operate in various business environments.
The Increasingly Strict Regulatory Trends are Risks that Businesses Must Face
“Starting a business means choosing something that won’t change in the future and then improving upon it,” believes Andy Lee. From an international perspective, digital advancements won’t reverse; technology can help us solve specific issues like information security. In the future, hacker attacks will become more frequent, systems more complex, and the issues of data storage and personal information protection will gain even more importance. Legal restrictions across countries will become more stringent, and businesses will face an increasing array of risks.
Andy Lee emphasizes that identity verification requires the integration of numerous technologies, including image and text encryption in network transmission, as well as the security of the app itself. Given their background in cybersecurity, they also pay special attention to product compliance, having obtained international certifications such as ISO27001 and ISO27701.
Participating in FIDO Alliance and Continually Enhancing Development Capabilities
Andy Lee is confident in the product technology offered by Authme. Approximately two-thirds of the team members are technical R&D personnel, continuously focusing on enhancing identity verification technology and user experience. In addition to regularly discussing newly published top-tier journal articles and timely incorporating the latest research findings into internal tests, technologies like federated learning and Edge AI are key areas of focus for the team. They are committed to ongoing research in cutting-edge knowledge to protect privacy and prevent identity fraud.
Furthermore, Authme actively participates in discussions on international organization standards and is a working group member of the FIDO (Fast IDentity Online) Alliance. “It’s a thankless job,” says Andy Lee. Not only are the meetings often held at midnight, but since the participants include representatives from various governments and corporations, reaching a consensus is a long and arduous process, often taking over a year to produce a preliminary draft. As such, few private companies are willing to devote the necessary effort to this task, but he believes that Taiwan’s industry needs more opportunities to participate in international dialogues.
