The Defense Advanced Research Projects Agency has completed the first real-world flight of an F-16 fighter jet controlled entirely by artificial intelligence, using the VENOM Autonomy Kit with human oversight capability standing by.
The milestone flight, conducted in late July and disclosed in early August, represents one of the most significant advances in autonomous military aviation since the advent of unmanned aerial vehicles. The VENOM system, short for Viper Experimentation and Next-gen Operations Model, allowed the AI to manage flight controls, tactical maneuvering, and real-time decision-making during a live sortie, while a safety pilot remained in the cockpit with the ability to assume manual command at any moment.
DARPA officials said the test was designed to evaluate how well machine-learning systems could handle the dynamic, high-stakes environment of fighter operations, where split-second decisions can determine mission success or failure. The AI was not following a pre-programmed script; instead, it responded to changing conditions in the airspace, adjusting altitude, speed, and heading based on sensor inputs and mission objectives. The agency reported that the system performed within acceptable safety parameters throughout the flight envelope tested.
The achievement builds on years of work by DARPA's Air Combat Evolution program, which has used simulated dogfights to train AI agents against human pilots. In those trials, AI systems repeatedly outperformed experienced fighter pilots in within-visual-range combat scenarios. Transitioning from simulation to a real aircraft, however, introduced engineering challenges related to sensor fidelity, actuator latency, and the physical stresses placed on the airframe during aggressive maneuvering.
Military analysts said the successful flight is likely to accelerate Pentagon discussions about the role of autonomous systems in future combat aircraft. The Air Force has already begun exploring crew-optional platforms under its Next Generation Air Dominance program, and the VENOM demonstration provides empirical evidence that AI can safely operate a manned fighter in real airspace. Questions remain about how autonomous systems will interact with human pilots in mixed formations, and what rules of engagement would govern an AI-controlled aircraft in a contested environment.
Beyond the military implications, the test highlights the broader trajectory of artificial intelligence toward handling high-consequence physical tasks in unpredictable environments. The same sensor fusion and real-time reasoning capabilities being developed for combat aviation have parallels in autonomous vehicles, industrial robotics, and disaster response systems. DARPA said additional flight tests are scheduled for the remainder of 2026, with progressively more complex mission profiles planned as the VENOM system matures.