Smart Engines has developed an updated "Sherlock 3o" system to combat digital fraud with fake documents. The solution detects fake images created by modern generative models such as NanoBanana, ChatGPT, Grok, Qwen, Midjourney, Stable Diffusion, Flux, and others.
The technology is designed for companies required to identify customers and combat fraud. The updated solution checks documents against 600 criteria and is used in banks and microfinance organizations.
According to the European Parliament, the number of deepfakes is growing exponentially, and the economic damage from them, according to Deloitte estimates, could reach $40 billion by 2027. In Russia, in 2026, with the participation of the Ministry of Digital Development, a interdepartmental working group was created to counter the illegal use of deepfake technologies.
This is the third version of the "Sherlock" anti-fraud system developed by Smart Engines scientists. "Sherlock 3o" is based on the analysis of micro-fragments of the image, identifying statistical anomalies and artifacts. The system also identifies collages with documents and assesses the quality of the image.
Currently, the "Sherlock" solution is used in international airports, banks, MFIs, the Federal Tax Service of Russia, and notary offices. With the help of the Smart Engines system, the Zaymer service has prevented more than 10,000 attempts to take out a loan using other people's passports using the system in a year.