UAE’s Fi 7 model detects malicious software with 94.7 percent accuracy: a sovereign standard facing 800,000 daily attacks

The UAE Government Cybersecurity Council announced an artificial-intelligence model dedicated to detecting malicious software and files, called "Fi 7", achieving an overall accuracy of 94.7 percent after being trained on a database of more than 200,000 records. The announcement came during the GISEC 2026 conference in Dubai, where it was presented by the UAE’s Chief Cybersecurity Officer, Dr. Mohammed Al-Kuwaiti, who described it as a practical tool aimed at handling a flow of threats that exceeds the capabilities of manual inspection and traditional signature-based matching systems across the nation’s critical infrastructure.

The model focuses on a specific niche limited to detecting and classifying malicious files rather than serving as a general security assistant, a domain where machine-learning algorithms can leverage labeled databases of malicious and benign samples and later test them against data the system has not previously seen. In evaluation, the model recorded 94.5 percent on an independent metric designed specifically to measure the quality of malware detection.From my position monitoring sovereign AI developments, I see the separation between overall accuracy and the detection-quality metric as a practical attempt to demonstrate the model’s ability to avoid false alerts that drain security-response teams.

The need to build and train the model locally stems from the volume of digital threats disclosed by Dr. Mohammed Al-Kuwaiti, with the UAE’s national infrastructure and strategic assets facing more than 800,000 cyber attacks per day. At such a high volume, relying on human analysts to review every suspicious file is infeasible, making automated triage an operational necessity. The announcement coincided with a 20 percent year-on-year growth of the GISEC 2026 conference, which brings together over 25,000 experts, more than 750 exhibiting brands and 350 speakers from 180 countries.

The launch of "Fi 7" falls within an international shift noted in the World Economic Forum’s Global Cybersecurity Outlook 2026, which found that 94 percent of organizations expect artificial intelligence to be the biggest driver of change in cyber defence, while 87 percent see its associated vulnerabilities as the fastest-growing threat. Although the Cybersecurity Council affirmed that its model outperforms other commercial detection systems in benchmark tests, details of the data sets and the false-alert rate have not yet been published in peer-reviewed studies, a pattern common to government initiatives that release detailed methodological frameworks only after implementation steps.

For enterprises and firms across the Gulf, Egypt and the broader Arab region, the model is not offered as a commercial off-the-shelf product, but rather sets a realistic baseline that reshapes market evaluation standards.When a sovereign entity announces a defensive model built on hundreds of thousands of records with an accuracy hovering around 95 percent, that figure becomes the benchmark against which security-solution providers negotiate and are required to exceed.This reality also forces a recalibration of the skills demanded of young security personnel in the region, shifting the focus to managing and auditing the outputs of intelligent models rather than spending time on initial manual review.

The approach primarily impacts sectors tied to critical digital infrastructure such as banking, aviation, ports, energy and telecommunications, whose security operation centers are directly linked to official cyber agencies. If the capabilities of "Fi 7" are incorporated into shared threat-intelligence channels or its results adopted as a regulatory standard demanded by oversight bodies, the project would gradually shift from an internal government tool to a compliance procedure against which major regional institutions measure their security spending and defensive architecture.