Biometric and document-based signals that are presented in two-dimensional formats are more difficult to evaluate under their digital format. It becomes harder to recognize the signs of manipulation. Furthermore, successful fraud is easily scalable online. It is very tempting for criminals to commit fraud. We see more and more sophistication and a growing business in the fraud of documents. Criminals are honing their techniques and sharing data online. We expect to witness an increase in fraud on documents as counterfeit ID cards are produced and then laminated. The problem of ID card fraud is becoming more difficult to solve. At the same time, disruptions and digitalization due to the pandemic mean that there are more opportunities for new fraudsters — motivated not only by luck but also by genuine financial necessity. With a few sophisticated, high-impact attacks, in addition to more naive attacks, Businesses are victims of attacks from both sides and must be aware of both. The positive side is that authentication services such as ClearDil use malicious attacks to feed data set used in the learning process of the models. Machine learning methods, which are constantly learning, improve the performances of the ClearDil API for KYC verification.