Hala Systems Sentry Warned Syrians of Airstrikes Before They Hit

An airstrike gives civilians almost no time to react. A jet traveling at hundreds of kilometers per hour can cross from distant airspace to a crowded neighborhood in minutes, and the people below often have no warning until the bombs fall. In Syria, where hospitals, schools, and markets were repeatedly hit during the civil war, a small technology nonprofit decided that those minutes did not have to be a mystery. Hala Systems built an AI system called Sentry that detected threatening aircraft, predicted where they were headed, and sent warnings to the people below before the planes arrived.

The results were measurable. During 2017 and 2018, Sentry covered nine Syrian governorates, reached over 2.3 million people, and was associated with an estimated 20 to 30 percent reduction in casualty rates in areas under heavy bombardment. Those numbers come from Nesta, the UK innovation foundation that studied the system as a landmark case of collective crisis intelligence.

From Plane Spotters to Prediction

Sentry began with volunteers. Residents who lived near airbases would watch for aircraft taking off and report what they saw through a smartphone app. These human sentries were the first data source, and they were remarkably effective, but their reports alone could not tell anyone where a plane was going or how fast it would arrive.

Hala Systems layered additional sources on top. Acoustic sensors placed in conflict zones detected the distinctive sound of military aircraft engines. Machine learning algorithms scraped social and local media for corroborating information about planes and flight paths. An AI model trained on historical attack data then fused all of these inputs, validated them against one another, and estimated the likely target and timing of a strike by comparing the current situation to past attacks.

The fusion step is what made Sentry trustworthy. Any single report could be wrong, delayed, or deliberately false. By demanding agreement across independent sensor types before issuing a warning, the AI reduced false alarms while keeping the system fast enough to matter. According to Hala Systems, at optimal range Sentry could identify threatening aircraft roughly 95 percent of the time.

Five Minutes That Save Lives

Once Sentry predicted a threat, the warning raced ahead of the aircraft through every channel available. Smartphone notifications and chatbots reached people with phones and internet. Television and radio broadcasts spread the alert further. On the ground, a network of remotely triggered air raid sirens, buzzers, and warning lights mounted in public spaces activated automatically. Direct messages went to hospitals, schools, and civil society organizations so that doctors could brace and families could move.

Civilians typically received five to ten minutes of warning. That is not enough time to evacuate a city, but it is enough time to get to a basement, move away from windows, gather children, and reach shelter. In the grim arithmetic of aerial bombardment, minutes are the difference between a strike on an occupied building and a strike on an empty one.

CBS News reported in 2018 that the system's alerts were believed to reach more than 2 million Syrians. The Nesta case study put the documented reach at over 2.3 million people across nine governorates during the 2017 to 2018 period, with an estimated 20 to 30 percent reduction in casualty rates where bombardment was heaviest.

Accountability Built In

Sentry did something else that early warning systems rarely do. It remembered. Reports and evidence collected through the system were stored immutably using blockchain technology, creating a tamper resistant record of attacks that could later support accountability efforts. Hala Systems has processed tens of thousands of pieces of digital evidence, and the company's documentation has supported investigations into attacks on civilians.

The recognition has been unusual for a conflict technology. The Smithsonian's National Air and Space Museum acquired three Sentry artifacts, a red warning light, a communications relay device, and an acoustic sensor, for permanent display, treating this civilian protection system as a piece of aerospace history alongside the aircraft it was built to detect.

Why This Model Matters Beyond Syria

The deeper innovation in Sentry is not any single sensor. It is the demonstration that scarce, messy, human gathered data from a war zone can be fused and validated by AI into something reliable enough to act on in seconds. Conflict zones are among the hardest environments on earth for data collection, and conventional monitoring systems often fail there entirely.

Hala Systems has since carried the model forward, adapting its early warning and violence monitoring work to new conflicts and to ceasefire monitoring, providing time critical information to civilians, first responders, and humanitarian organizations who need to make life or death decisions about where to deliver aid and where to take cover. The company describes Sentry as the only data driven early warning system of its kind serving civilians and first responders directly.

There is a lesson here that extends well beyond war. When people say AI for good, they usually mean models that spot disease or predict famine. Sentry is a reminder that the category also includes systems that give ordinary people a few more minutes of warning, a chance to grab their children and run, in the moments when nothing else can help.