Dubai’s New AI Model Can Spot Deepfake Videos with 91% Accuracy

Dubai is preparing to launch SARAAB, a government-developed AI model designed to detect deepfake videos. Developed by the Dubai Electronic Security Centre, the system can identify manipulated sections and highlight altered facial areas through heat maps.
Dubai’s New AI Model Can Spot Deepfake Videos with 91% Accuracy
Written By:
Soham Halder
Reviewed By:
Manisha Sharma
Published on

Dubai is preparing to launch an artificial intelligence model designed to identify deepfake videos, with the system expected to become available by the end of 2026. Developed by the Dubai Electronic Security Centre (DESC), the model is called SARAAB and is designed to identify manipulated sections of video and show where facial alterations have been made.

The announcement was made during GISEC Global 2026, the three-day cybersecurity exhibition and conference taking place from September 16 to 18 at Dubai Exhibition Centre in Expo City. DESC said SARAAB has recorded an accuracy rate of 91% during its development.

SARAAB Uses Heat Maps to Flag Manipulation

Unlike systems that simply assess whether a video has been altered, SARAAB is designed to identify the specific facial areas affected by manipulation. It uses a heat map to highlight those portions, giving users a visual indication of where a deepfake may have been introduced.

Hassan Majid Alkhazraji, Senior Artificial Intelligence Executive at DESC, said that the model can analyze an entire video rather than examining only selected frames. If manipulation appears only toward the end of a longer recording, the system can also identify that section.

“If it was a two-minute video, we would analyze it for the whole of two minutes. Maybe the last 30 seconds are deepfaked. So, that will also be analyzed as well with the model,” Alkhazraji said.

Government-Developed Model to be Open Source

DESC plans to make SARAAB available through Hugging Face, an open-source platform where users can access models and, depending on their licensing, examine or modify them. According to Alkhazraji, SARAAB is the first AI model developed by a government entity in the Arab region for this purpose. 

DESC reportedly spent around six months developing the system, following concerns from government bodies about deepfakes used in fraudulent advertisements and other deceptive content.

The move comes as AI-generated video becomes increasingly difficult for ordinary users to distinguish from authentic footage. Earlier this year, UAE authorities also urged residents to verify suspicious videos and images before sharing them, particularly content appearing to involve public figures or official institutions.

“Dubai continues to establish an advanced model for cybersecurity, built on anticipating change, developing national capabilities, and transforming challenges into opportunities for innovation and progress,” said Yousuf Hamad Al Shaibani, Chief Executive of the Dubai Electronic Security Centre.

Also Read: Cybersecurity, Resilience in Focus as UAE Central Bank Revamps Operational Risk Rules

Dubai Adds New Cybersecurity Initiatives

SARAAB was announced alongside other cybersecurity initiatives at GISEC Global 2026. DESC introduced the Dubai Cyber Skills Framework, which establishes a common reference for qualifications, certifications, training and career progression in cybersecurity.

The centre also launched the ISR Auditor Certification Program to help information-security professionals and auditors assess compliance with Dubai's Information Security Regulation through defined standards and methodologies.

With SARAAB, Dubai is adding another layer to its efforts to address the growing challenge of digitally manipulated content while making locally developed AI technology available to a wider user community.

Analytics Insight UAE: Top Tech News Website in UAE, Dubai & Middle East
www.analyticsinsight.ae