News Resource：Central News Agency
Translated by Authme
Anyone who has dealt with banks is familiar with the experience of filling out lengthy forms. When applying for a bank account online, you must upload a photo of your ID and enter personal information, but sometimes, if the photo isn’t taken correctly, you have to start over. Proving «you are who you say you are» in the digital world is both a challenge and an opportunity.
Authme, a startup founded by a group of white-hat hackers, targets the digital identity verification market. Based on artificial intelligence technology, they offer a one-stop solution that can even detect facial skin texture and blood flow, cracking recent deepfake technology.
Andy Lee, co-founder and CEO of Authme, ventured into a cryptocurrency exchange in 2017. From this, he saw the business opportunity in digital identity verification. In the three years since Authme was established, it has served brands like pure online bank LINE Bank and shared transportation vehicle WeMo Scooter.
Andy observed that many cryptocurrency exchanges were international platforms. When users registered, the platform required them to upload a selfie holding their ID as part of the Know Your Customer (KYC) verification process.
Andy explains that during the registration process, the exchange gives users a code, similar to a «One Time Password» (OTP). Users must submit a selfie during registration, which clearly shows at least three items: the user, the ID they are holding, and an A4 paper written with the exchange’s account number and the date.
This means the user must ensure these items are clearly presented in the photo, or it could affect the subsequent verification process. Users typically wait 3 days to receive notification of approval. «This might already involve dozens to hundreds of people constantly working on verification to process it that quickly,» says Andy. If there are doubts about the uploaded information, users might also receive a notification to resubmit documents after 3 days.
From his experience in cryptocurrency exchange entrepreneurship and industry, Andy saw two major pain points in identity verification: First, the process was not user-friendly enough, leading to potential customers abandoning the effort and businesses losing them. Second, enterprises relying on manual verification not only incurred costs but also, with technological advancements, human eyes might not accurately detect identity fraud cases.
In 2019, the team founded Authme, based on artificial intelligence and deep learning. They combined patent NFC chip verification, document checking, facial recognition, and live detection technologies to provide a one-stop digital identity verification service, which could also detect if a facial image has been altered by deepfake technology.
Andy exemplifies that Authme’s technology, integrated into financial industry’s online banking apps, mainly includes two stages of verification. The first stage involves detecting the user’s documents and verifying whether the documents match the actual person. With the rapid increase in fraud incidents like deepfakes, Authme also examines through biometric recognition. For instance, the forehead skin is thinner, so during facial detection, skin texture and blood flow are observed, and models are used to verify if it’s a real person, avoiding deepfake fraud. The second stage involves secondary detection to prevent image alteration during transmission to the backend.
According to statistics from VMware in the USA, over 60% of cybersecurity personnel surveyed had experienced security incidents involving deepfakes. Andy states that technology is neutral; deepfake technology can be used in movie production but, on the other hand, also in emerging fraud incidents. Although currently, judgments can be made through facial edges, hairline blending, or biometric recognition, future challenges will inevitably arise from increasingly realistic deepfake technology.
Despite the rapid changes in technology, Andy, with his white-hat hacker background, lights up when talking about technology. Authme’s current team size is about 30 people, with approximately two-thirds being R&D personnel. Each week, the team discusses newly published top-tier journal articles, researching and timely integrating findings into internal testing to ensure they stay at the forefront of technology.
Regarding concerns about competition from major international companies, Andy frankly says that although international companies have vast data on facial recognition, Authme focuses on facial fraud detection, which is a different aspect. Plus, Authme specializes in Asian facial recognition and has built its own moat. «So far, we haven’t lost to any foreign companies.»
