SecurityBrief Australia - Technology news for CISOs & cybersecurity decision-makers
Story image
Sumsub launches 'For Fake’s Sake' to detect deepfakes and synthetic fraud
Mon, 23rd Oct 2023

Sumsub, a full-cycle verification platform, has announced the launch of 'For Fake's Sake', a pioneering set of models purposed to detect deepfakes and synthetic fraud within visual assets. This innovation enables users to estimate the likelihood of an uploaded image having been artificially created.

The announcement of 'For Fake's Sake' comes at a crucial time, as there has been a significant increase in deepfake cases globally. The Asia-Pacific region, in particular, has observed a considerable uptick in AI-generated identity fraud instances, with notable spikes in countries like Australia (1300%), Vietnam (1400%), and Japan (2300%).

'For Fake's Sake' represents the first publicly accessible tool developed by a verification provider. With this launch, Sumsub seeks to offer the AI community, a cohort including developers, AI researchers, and scientists, a platform upon which they can test, experiment and explore innovative countermeasures to the menace of deepfakes.

Sumsub's in-house AI/ML Research Lab is behind the development of 'For Fake's Sake', assembling four distinct machine learning models for deepfake and synthetic fraud detection. Upon using 'For Fake's Sake', users can understand how likely it is that an image has been artificially generated.

Following the AI community's guidelines, Sumsub furnishes detailed cards and comprehensive documentation outlining their AI models' datasets and performance metrics. Results from Sumsub's internal testing suggest that these AI models effectively recognise typical image alterations.

Analysis of Sumsub's most recent internal data reveals a sharp increase in deepfake cases worldwide during the first half of 2023 compared to the second half of 2022. Countries such as Australia, Vietnam, and Japan in the Asia-Pacific (APAC) region registered an alarming surge in AI-generated identity fraud instances. The number of deepfakes also grew substantially in Great Britain (84%), the U.S. (250%), Germany and Italy (over 300%), and France (500%).

Vyacheslav Zholudev, Co-Founder and CTO of Sumsub, says: "We've dedicated substantial efforts to combat the menace of deepfakes and the adverse impacts of synthetic fraud by constantly upgrading our existing solutions and developing new features, including the newly enhanced deepfake detector of our liveness tool." H

He explains that Sumsub is committed to offering customers top-quality products and playing an active role in maintaining a transparent public discourse on the multiple threats posed by deepfakes. This includes illicit online impersonation, personal data theft, misinformation spread, social engineering, and more.

Following this preliminary release, Sumsub plans to exploit the feedback from the AI research community to enhance the models' capabilities, allowing them to evolve alongside other AI-driven tools.

Pavel Goldman-Kalaydin, Head of AI/ML at Sumsub, cautioned that as AI technologies advance, there may be a necessity to apply watermarks to all synthetic images.

He highlights: "We're pleased to share our experimental synthetic image detection models as a benchmark for further development in the battle against AI-generated fraud. Our team is committed to continued development based on the latest trends, user feedback, and community insights."