In Russia, Smart Engines has presented a new system for remote identity verification — Smart ID Engine 2.5. This was reported by the company's press service.
The system is capable of recognizing and verifying the authenticity of any documents in photos, scans, and video streams. This will allow banks and other financial organizations to combat fraud.
The new version of the Smart ID Engine 2.5 system operates on the basis of two AI architectures — "Da Vinci" and "Sherlock", which perform different tasks. On average, both models require no more than a second to analyze one image.
The "Sherlock" neural network uses a scientific-deductive approach to verify the authenticity of documents, recognizing both simple and complex forgeries. "Sherlock" conducts 533 checks, divided into 13 groups. It analyzes the digital authenticity of the document, detects interference in the image, studies the sample in various light ranges, and checks the consistency of details.
Unlike similar systems, the new neural network clearly explains its decisions. This helps to strengthen the protection of information and funds from attackers.
The developers emphasize that the training of AI and its operation are conducted in full compliance with the law. Special sets of graphic data were created for training the systems, and real documents are stored only by the customer and are not transferred to third parties.
Earlier, www1.ru reported that a Russian development will help to identify fake voices.
Read materials on the topic:
AI fraudsters extort money using fabricated voices of the SUSU leadership
Yandex neural networks are seven times faster at finding fraudsters
Telegram hacking with AI - how fraudsters in Russia use your voice and passport
