The US Air Force Just Let AI Fly a Real F-16
On July 16, 2026, at Eglin Air Force Base in Florida, a modified F-16 fighter jet took to the air — but the aircraft wasn’t being flown by the pilot sitting in the cockpit. It was being flown by an AI system. The pilot stayed on board the whole time, monitoring the aircraft, with a single switch within reach: flip it, and control snaps straight back to human hands. This is the latest milestone in DARPA and the US Air Force’s VENOM program, and the first time “letting AI fly a real fighter jet” has happened outside a simulator.
What VENOM actually is
VENOM stands for Viper Experimentation and Next-generation Operations Model, a joint DARPA-Air Force program testing autonomous flight. Its core piece of hardware is the VENOM Autonomy Kit (VAK), which automates flight controls and sensor systems without touching the F-16’s core software — in other words, the AI runs as a bolt-on layer on top of an existing fighter jet, not a full rewrite of the aircraft’s brain. VENOM builds on another DARPA program called Air Combat Evolution (ACE), whose flagship platform is the X-62A VISTA — a legendary test aircraft that’s been flying for more than three decades.
This didn’t come out of nowhere
VENOM’s credibility rests on a track record VISTA already built. Between December 2022 and September 2023, the ACE program flew 21 test flights on the X-62A. Then in April 2024, an AI-controlled X-62A VISTA actually dogfighted a human-piloted F-16 over Edwards Air Force Base in California: the test started with defensive maneuvers to establish safety margins, then escalated to offensive, nose-to-nose engagements where the two jets closed to within 2,000 feet of each other at 1,200 miles per hour — and at no point during the entire engagement did the test pilot need to hit the safety switch. The achievement was later a finalist for the Collier Trophy, one of aerospace’s highest honors. Then-Secretary of the Air Force Frank Kendall put it this way: “The potential for autonomous air-to-air combat has been imaginable for decades, but the reality has remained a distant dream up until now.”
Flip the switch, hand over control
This round of VENOM testing ran in two phases. In June, the team first had a human pilot fly the modified F-16 to confirm all the new hardware and software upgrades worked safely in the air — checking avionics integration, flight control response, and overall airworthiness. It wasn’t until July that the program formally moved into having an onboard AI agent take over portions of the flight, with the human pilot continuously monitoring aircraft performance and mission execution throughout. DARPA describes the human-machine handoff interface as “novel,” emphasizing that pilots can switch between traditional and AI control “with the flip of a switch,” without compromising safety.
What the officials are saying
Brig. Gen. James “Fangs” Valpiani, the DARPA program manager leading the effort, said the breakthrough flight tests “advance the infrastructure needed to develop trusted, autonomous air combat capabilities” — but he also acknowledged: “Many hard questions remain concerning the performance and trustworthiness of combat AI.” Lt. Col. Patrick “Dice” Highland, the incoming program manager, laid out a more ambitious vision: “We now have the opportunity to create dominant autonomy for beyond-visual-range, multi-ship combat.”
What comes next
VENOM is set to fold into a new DARPA effort called AI Reinforcements (AIR), with the goal of having the whole VENOM fleet test multiple AI agents simultaneously in live flight, scaling up toward more complex multi-ship missions — eventually letting a single human pilot command an entire team of autonomous, uncrewed fighters. That directly feeds into the Air Force’s broader push for what it calls Collaborative Combat Aircraft. The F-16 remains one of the US military’s active frontline fighters today, which means VENOM isn’t testing a lab concept car — it’s testing a jet that could genuinely end up flying into combat.
It’s worth noting that even Valpiani himself, the man leading this program, isn’t overselling it — he’s on record acknowledging that combat AI’s performance and trustworthiness still raise “many hard questions.” That’s precisely why VENOM deliberately kept a physical toggle switch in the cockpit: until AI can prove it performs reliably in high-stakes air combat scenarios, letting a human pilot reclaim control with one flip is the only safety guarantee this system can currently offer the outside world. How Valpiani’s “hard questions” ultimately get answered will come down to how the next rounds of flight testing go.
