Deepfake Detection AI: SARAAB Model Launched in Dubai Meta

 Deepfake Detection AI: Dubai Unveils SARAAB Open-Source Model

Addressing synthetic media requires moving beyond isolated software solutions toward an integrated ecosystem combining technical tools, workforce development, and institutional governance.

At GISEC Global 2026, the Dubai Electronic Security Centre (DESC) introduced three major cybersecurity initiatives aimed at strengthening digital resilience. As reported by Gulf News, the central initiative is SARAAB, an open-source deepfake detection AI developed entirely by an Emirati team to enable collaborative testing among researchers and engineers. 

According to details published by Khaleej Times, the SARAAB AI model achieves a reported 91% accuracy rate, utilizing heat-map analysis across full video durations, with deployment scheduled on Hugging Face by the end of 2026. Furthermore, Emirati Times highlights that DESC simultaneously launched the Dubai Cyber Skills Framework alongside the ISR Auditor Certification Program to establish standardized training pathways and compliance methods across Dubai cybersecurity operations.

The release of SARAAB demonstrates a shift toward open-source framework adoption within government-led security initiatives. By pairing an AI detection tool with standardized auditor qualifications and competency benchmarks, public institutions can address synthetic media threats through both technical verification and regulatory accountability.

Can open-source government models effectively establish global standards for digital trust?

FAQs

What is SARAAB?

SARAAB is an open-source AI model developed by DESC designed to detect deepfake videos using heat-map analysis.

Where will the SARAAB AI model be hosted?

The model will be deployed on developer platforms, including Hugging Face, for public research and development.