Key Takeaways
- Dubai’s Electronic Security Centre has launched SARAAB, an AI model for detecting deepfake videos.
- SARAAB has a 91% accuracy rate and is the first such tool in the Arab world developed by a government agency.
- The model will be released as open-source, allowing researchers and developers to study and improve it.
Dubai’s Electronic Security Centre has unveiled SARAAB, an artificial intelligence model designed to detect deepfake videos with 91% accuracy. The tool was introduced at GISEC Global 2026 in Dubai.
Developed by an Emirati team in just six months, SARAAB is the first deepfake detection model in the Arab world built entirely by a government agency. It aims to identify digitally manipulated content and counter impersonation threats in both public and private sectors.
Unlike other tools that analyze only single images, SARAAB scans entire video clips and generates heat maps with moving hotspots to identify manipulated frames. This comprehensive approach allows it to detect even short clips containing altered frames within otherwise genuine footage.
The Electronic Security Centre stated that SARAAB will be released as an open-source version on platforms like Hugging Face. This move is intended to allow researchers, developers, and engineers to study the source code, test performance, and build customised versions.
The centre believes that SARAAB can help protect digital media, strengthen cybersecurity, and improve trust in online content. By making the tool available to the broader community, the Electronic Security Centre hopes to enhance its effectiveness and reach.
SARAAB’s introduction marks a significant step in the fight against deepfake videos, which have become increasingly prevalent and sophisticated. The tool’s ability to scan entire video clips and identify manipulated frames makes it a valuable asset in the digital age.
The launch of SARAAB underscores Dubai’s commitment to advancing cybersecurity and digital security. As the tool becomes more widely available, it could play a crucial role in safeguarding online content and protecting against potential impersonation threats.
