Sumsub, a global verification platform, has announced the launch of its enhanced deepfake detection technology. The company has seen a significant increase in the number of deepfakes detected in the first quarter of 2023 compared to all of 2022 combined. In response to this growing threat of identity fraud, Sumsub has made major updates to its in-house liveness solution, which is a facial biometrics tool and an integral part of the KYC flow for businesses.
The deepfake fraud detector now integrates into Sumsub’s liveness solution and has been enhanced to align with the latest AI technology. This ensures that no fraudster can pass the check and helps combat the advancing malicious technologies used by fraudsters.
“We noticed the emerging deepfake trend long ago and started building solutions to fight this rapidly evolving type of fraud,” said Vyacheslav Zholudev, co-founder and CTO of Sumsub. “As an anti-fraud provider, Sumsub strives to stay ahead of the game. We continuously develop innovative fraud detection solutions to combat the advancing malicious technologies and prevent potential damage to digital platforms, people, and communities. Our new advanced deepfake detector reinforces our core KYC product and commitment to further enhancing the AI-driven technologies built into our anti-fraud solution.”
Sumsub’s proprietary liveness technology easily detects spoofing attempts while authenticating real users in seconds. The system creates a 3D FaceMap to ensure users are physically present, which is continuously referenced for authorizing all future actions such as transactions and logins.
All businesses operating online are vulnerable to deepfakes, with fintech, payments, crypto, and gambling platforms being at particularly high risk. Sumsub’s internal identity fraud statistics show that deepfakes are continuously replacing ‘simple’ kinds of identity fraud, such as document printouts.
The data revealed that in 2022, the leader in deepfake fraud was Spain, accounting for 49.7% of all global cases, followed by Great Britain (9.3%) and the U.S. (4.2%). In the first quarter of 2023, the most deepfakes came from Great Britain and Spain, accounting for 11.8% and 11.2% of global deepfake fraud, respectively. Germany (6.7%) and the Netherlands (4.7%) also had significant cases. The U.S. held fifth place, representing 4.3% of global deepfake fraud cases.
The proportion of deepfakes among all fraud types has significantly increased in Canada by 4,500%, the U.S. by 1,200%, Germany by 407%, and the UK by 392% from 2022 to the first quarter of 2023. In the last quarter, a high proportion of deepfakes among all fraud types was also noticed in Australia (5.3%), Argentina (5.1%), and China (4.9%).
“As we face the ever-evolving trends in generative AI and AI-driven fraud, we anticipate AI technologies to quickly develop, and more sophisticated fraud to emerge, affecting even more industries. The use of synthetic fraud is rising at an alarming rate and leading to the rapid spread of misinformation, as recently seen with the fake images of the explosion at the Pentagon, resulting in significant media hype. As AI advances, more tools become available to fraudsters, and Sumsub is here to fight all kinds of synthetic fraud. Our team is working on a solution to detect a large scope of AI-generated fraud beyond deepfakes,” says Pavel Goldman-Kalaydin, Head of AI/ML at Sumsub.
To learn more about Sumsub’s deepfake detection technology, visit https://sumsub.com/blog/liveness-and-deepfake-detection/.
Sumsub is a full-cycle verification platform that secures the entire user journey. With Sumsub’s customizable KYC, KYB, transaction monitoring, and fraud prevention solutions, businesses can orchestrate their verification process, welcome more customers worldwide, meet compliance requirements, reduce costs, and protect their business. The company has over 2,000 clients across various industries, including fintech, crypto, transportation, trading, e-commerce, and gaming. Some of its clients include Binance, Mercuryo, Bybit, Huobi, Unlimint, DiDi, Poppy, and TransferGo.