Friday, December 6, 2024
12.4 C
Los Angeles

FATF Monitoring: Countries Addressing Strategic Deficiencies

Jurisdictions under Increased Monitoring by the FATF Countries...

Former Peruvian President Alejandro Toledo Sentenced to 20+ Years in Odebrecht Bribery Scandal

Former Peruvian President Alejandro Toledo has been...

Ex-Mexican Security Chief Sentenced for Bribery and Aiding Sinaloa Cartel’s Drug Trafficking

Genaro Garcia Luna, Mexico's former Secretary of...

Sumsub Launches Open-Source Deepfake Detection Models to Combat Synthetic Fraud

AI/MLSumsub Launches Open-Source Deepfake Detection Models to Combat Synthetic Fraud

Sumsub, a full-cycle verification platform, has unveiled “For Fake’s Sake,” a set of machine learning models designed to detect deepfakes and synthetic fraud in visual assets. This marks the first such solution developed by a verification provider and made publicly available for download and usage. Sumsub’s goal is to provide the AI community, including developers, researchers, and scientists, with a platform to experiment with and develop innovative ways to combat the growing threat of deepfakes.

The in-house AI/ML Research Lab at Sumsub created four distinct machine learning-driven models for deepfake and synthetic fraud detection. These models estimate the likelihood that an uploaded image was artificially created. Sumsub adheres to AI community guidelines by offering comprehensive documentation describing the datasets and performance metrics of their AI models. Internal testing suggests that these models are effective at accurately detecting typical image alterations. When used alongside other content analysis methods, AI-generated images can be more confidently recognized.

Sumsub’s internal data for the first half of 2023 indicates a significant increase in deepfake cases worldwide compared to the second half of 2022. The most substantial surge in AI-generated identity fraud cases was observed in Asia-Pacific (APAC) countries, such as Australia (1300%), Vietnam (1400%), and Japan (2300%). The number of deepfakes also increased by 84% in Great Britain, 250% in the U.S., over 300% in Germany and Italy, and 500% in France.

Following this release, Sumsub plans to use feedback from the AI research community to further refine and expand the models’ capabilities, ensuring the platform remains adaptable and aligned with other AI-driven tools.

By FCCT Editorial Team

Disclaimer: The views expressed in this article are independent views solely of the author(s) expressed in their private capacity.

Check out our other content

Ad


Check out other tags:

Most Popular Articles