Sumsub launches ML models for detecting deepfakes
“As AI technologies advance, we foresee a tightening of regulations governing their use. However, fraudsters will continually seek ways to overcome regulations, especially leading up to the 2024 U.S. presidential elections.”
In a move that marks a significant leap forward in the fight against deepfake technology and synthetic fraud, Sumsub, a leading verification platform, has launched “For Fake’s Sake,” a set of machine-learning models designed to detect manipulated visual assets.
This groundbreaking tool is the first of its kind developed by a verification service provider and is freely available for public use.
Deepfakes, a threat to individual privacy and societal trust
Deepfakes—convincing fake videos and images generated by artificial intelligence—pose a growing threat to both individual privacy and broader societal trust.
Sumsub, which specializes in identity verification, has taken this challenge head-on by extending its in-house AI and machine learning research to develop For Fake’s Sake. The tool employs four unique machine learning models that aim to unmask AI-generated fake images, thereby empowering people to safely interact with digital content.
Vyacheslav Zholudev, co-founder and CTO of Sumsub, commented, “Our open-source tool represents just one of the many strides Sumsub is taking, apart from product development, towards fulfilling our mission of empowering a safe and inclusive digital future—for companies and the wider public’s free use alike.”
Deepfakes on the rise across the globe
Recent internal data from Sumsub underscores the urgency of combating deepfakes. The first half of 2023 saw an alarming rise in deepfake cases globally: an 84% increase in Great Britain, 250% in the U.S., over 300% in Germany and Italy, 500% in France, and staggering surges in Australia (1300%), Vietnam (1400%), and Japan (2300%).
Pavel Goldman-Kalaydin, Head of AI/ML at Sumsub, warned, “As AI technologies advance, we foresee a tightening of regulations governing their use. However, fraudsters will continually seek ways to overcome regulations, especially leading up to the 2024 U.S. presidential elections.”
Sumsub’s For Fake’s Sake allows users to assess the likelihood that an image is artificially created. Following community guidelines, it also provides detailed metrics and dataset descriptions for its models.
While internal testing has demonstrated its efficacy, the company plans to continually adapt and improve the tool based on feedback from the AI-research community.
In an era where digital trust is under siege, Sumsub’s For Fake’s Sake sets a new benchmark for combating synthetic fraud. As the threat landscape evolves, the tool stands as a public safeguard, marking a seminal contribution to an ongoing, industry-wide effort to restore integrity to the digital realm.