Harmattan AI’s new IRIFI drone is a claim before it is a proven capability. The French startup says the one-way system can travel up to 2,000 kilometres, navigate without satellite positioning and complete an approved mission without a human operator controlling it during flight. Independent flight-test evidence, cost data, payload details and production maturity have not been made public.
But the significance lies beyond one aircraft. For years, the defence against long-range drones has relied partly on disrupting satellite navigation and communications. Jam GPS. Break the command link. Force the aircraft to lose its route, lose its operator or lose its target.
IRIFI is designed around the assumption that those methods may no longer be enough. Harmattan says its system uses onboard sensing and processing, multi-source navigation and visual terminal guidance to continue operating when GNSS signals or communications are degraded. Its stated approach is to have a human commander designate authorised targets and set mission parameters before launch, after which the vehicle completes the mission autonomously within those boundaries. That does not mean GPS jamming has become irrelevant. It means it may become only one layer of defence.
A drone that can navigate without GPS still must be detected, identified, tracked and stopped. Air-defence systems may use radar, electro-optical sensors, acoustic detection, radio-frequency monitoring and other surveillance tools. Airports, power facilities, ports and city authorities need systems that can distinguish an authorised aircraft from a potential threat without creating unacceptable risks to ordinary aviation.
That is the civil-airspace problem. Traditional aviation depends on cooperative systems. Aircraft use navigation signals, communicate with air traffic control and, in many cases, broadcast identifying information. A long-range autonomous weapon is designed to operate when those assumptions fail or are deliberately absent. It can therefore complicate the task of protecting civilian airspace, particularly where a threat approaches at low altitude or through areas crowded with ordinary traffic.
The danger is not simply collision. It is confusion. An airport or air-navigation provider may have seconds or minutes to decide whether an unidentified object is a small civilian drone, a faulty aircraft, a legitimate government operation or a hostile system. A wrong decision can close airspace unnecessarily. A delayed decision can put aircraft, passengers and airport workers at risk.
The announcement also tests existing accountability rules. Harmattan says humans approve the targets and define the mission limits. That is an important claimed safeguard, but it shifts attention to the chain of authorisation who validates target intelligence, who establishes the geographic and time boundaries, what abort options exist, and how an investigation would reconstruct an error after a system completes a mission without a live operator.
Those questions are legal as well as technical. International humanitarian law requires distinction, proportionality and precaution. A weapon with greater autonomy does not remove human responsibility. It makes the record of human decisions more important.
Export controls present another gap. Long-range autonomous strike systems combine capabilities that are often regulated separately: unmanned aircraft, guidance technology, software, optics, communications equipment and military payloads. An export licence for an airframe does not settle how governments should evaluate autonomy, range, production partnerships or downstream transfer.
Morocco and Harmattan AI have signed a letter of intent to develop IRIFI as a sovereign Moroccan system, but the programme remains at an early stage. The company has said it expects production by the end of 2027.
The promise remains a promise till verified a 2,000km GPS-independent flight, operational accuracy in a contested environment or repeatable mass production.
Still, the direction is clear. Drone warfare is moving away from the assumption that an aircraft must remain connected, continuously piloted and dependent on satellite signals.
The next security question is no longer only how to jam the drone. It is whether civilian airspace, critical infrastructure and border defences can detect and defeat a system built to keep moving after the signal disappears.