A U.S. Air Force F-16 has flown with an artificial intelligence agent autonomously in command of the aircraft, a milestone that moves machine-piloted air combat a major step closer to the operational fleet. The flights, conducted under a joint program run by the Defense Advanced Research Projects Agency and the Air Force, mark the first time an AI has taken the controls of a standard, service-model F-16 rather than the one-of-a-kind experimental jet used in earlier tests. A human pilot remained in the cockpit throughout, able to switch control back with the flip of a switch.
The VENOM program and its first flights
The aircraft is one of a group of F-16s converted into autonomous-capable testbeds under the Viper Experimentation and Next-generation Operations Model, known as VENOM, a partnership between the Air Force and DARPA. In an announcement dated July 16, 2026, the agency said a VENOM-modified F-16 based at Eglin Air Force Base in Florida had begun in-air testing using an AI agent to autonomously control flight. The agency described the achievement as a demonstration of the United States’ ability to transform standard operational aircraft into platforms that can run cutting-edge combat AI.
Building on the X-62A VISTA
The new flights extend earlier work with the X-62A VISTA, a one-of-a-kind, heavily modified experimental jet that first proved an AI agent could autonomously fly a fighter through a dogfight. That aircraft was a purpose-built research platform, engineered to mimic the handling of other aircraft and to safely host aggressive maneuvering algorithms. The VENOM effort, part of DARPA’s broader Air Combat Evolution program, is significant precisely because it takes that capability off a singular testbed and onto ordinary fleet jets, a change that matters enormously for whether autonomy can ever be fielded at scale.
How the autonomy kit works
The modification, called the VENOM Autonomy Kit, was designed and integrated by multiple performers under the program. According to DARPA, it uses a novel interface with the aircraft’s flight controls and mission systems while leaving the jet’s core software unchanged, allowing a pilot to toggle between traditional human control and AI control instantly. That design keeps a person on the loop, monitoring the AI agent and ready to intervene, which the agency describes as essential for safe, reliable experimentation. Brigadier General James Valpiani, a DARPA program manager, said the tests advance the infrastructure needed to develop trusted, autonomous air-combat capabilities and create an efficient pipeline for building combat AI.
What comes next under the AIR program
DARPA said the VENOM fleet will serve as the cornerstone for the next phase of development under its Artificial Intelligence Reinforcements program, or AIR, which aims to test multiple AI agents in live-flight scenarios of increasing complexity. The longer-term goal is to let human pilots command and orchestrate teams of autonomous, uncrewed aircraft, capabilities that could feed into the Air Force’s Collaborative Combat Aircraft efforts to pair crewed jets with drone wingmen. Program officials framed the work as an early glimpse of how AI agents may begin actively transforming air warfare, particularly in beyond-visual-range engagements where the volume of information can overwhelm a human.
A key selling point of the VENOM design is that it converts standard F-16s without rewriting the jet’s core flight software, an approach the program describes as an efficient pipeline for developing combat AI. Rather than building one exquisite, expensive testbed, the effort aims to make many ordinary aircraft available for experimentation, so that competing AI agents can be flown, evaluated and improved rapidly. That emphasis on speed and repeatability reflects a wider Pentagon concern that autonomy must be developed quickly enough to keep pace with rival powers investing heavily in the same technology.
The questions that remain
Officials were candid that hard problems endure. The emerging threat environment for aerial combat is growing more complex, and DARPA leaders acknowledged that many questions remain about the performance and trustworthiness of combat AI in the chaos of real warfare. Keeping a pilot in the cockpit to supervise reflects how much testing lies ahead before any autonomous system could be trusted without direct human oversight. Trust is the recurring theme in the agency’s own descriptions of the work: an AI that flies well in a controlled test must still prove it behaves predictably when confronted with the confusion, deception and split-second stakes of combat. The flights represent a demonstration and a research step, not the deployment of an operational weapon, and the agency stressed that scaling up to multi-aircraft operations will require substantial further work. Even so, moving autonomous flight from a bespoke research jet onto a workhorse fighter signals that the technology is edging out of the laboratory and toward the aircraft the military actually flies.
This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.
More from Morning Overview
- Long-term use of common heartburn pills is tied to kidney and dementia risk
- A California supervolcano has bulged upward about two and a half feet since 1978
- The FTC is warning about a scam quietly draining thousands from victims
- The NSA is again telling phone owners to switch off one location setting
