The US Air Force’s X-62 VISTA experimental aircraft has demonstrated an artificial intelligence agent using live infrared sensor data to direct the aircraft towards an airborne target, the UK Defence Journal understands.
The test formed part of two accelerated programmes conducted at Edwards Air Force Base during April 2026. Known as HAVE HEAT and HAVE HOLIDAYS, the parallel campaigns examined autonomous control, open mission systems and safety measures intended to support further testing of artificial intelligence aboard the X-62.
During HAVE HEAT, the aircraft used data from a Lockheed Martin Legion Pod, an infrared search-and-track sensor designed to detect and track airborne targets through their heat signatures. The AI agent processed the live sensor feed and issued commands directing the X-62 to intercept a target in real time.
The activity linked sensing, onboard processing and aircraft control within a single test sequence. According to the US Air Force, the work was intended to support future trials involving AI-controlled sensors and multiple air vehicles.
Lieutenant Colonel Joshua Strafaccia, dean of faculty for research at the US Air Force Test Pilot School, said, as quoted in the news update: “HAVE HEAT represents a meaningful expansion of avionics capability towards integrated, AI-driven control of multiple sensors and air vehicles for mission autonomy. This bridges a capability gap and positions us to test advanced autonomy faster.”
A second programme, HAVE HOLIDAYS, ran alongside the interception trials. Engineers integrated and evaluated several modular technologies through the aircraft’s Enterprise Open Mission Systems Architecture computer, which provides a common framework for adding and testing new software and hardware.
The work included loading an autonomous agent developed by a third-party supplier rather than the aircraft’s principal contractor. Test teams also examined enhanced safety rules intended to prevent autonomous systems from moving beyond operational limits set by human users.
Additional computing chips were integrated for advanced sensor processing. The Air Force described the campaign as a system-of-systems test intended to reduce technical risk before the next stage of the aircraft’s mission-system upgrade programme.
Military personnel, government specialists and contractor engineers completed the associated ground testing within three months. HAVE HEAT and HAVE HOLIDAYS were then flown in parallel under compressed schedules, allowing the Test Pilot School to assess several technologies before the planned upgrade work continues.
The X-62 VISTA is operated by the US Air Force Test Pilot School as a dedicated research and flight-test aircraft. Its configurable systems allow researchers to install, evaluate and refine new control laws, autonomy software and human-machine teaming technologies without assigning the work to an operational combat aircraft.
Maintaining the aircraft within the Test Pilot School also allows students, engineers and researchers to work in a shared test environment. The Air Force said this arrangement supports continued development of AI systems while retaining human oversight and incorporating pilot input into the models being assessed.
The next phase of the Mission Systems Upgrade programme is expected to add the RTX PhantomStrike radar, expand onboard computing and sensor capacity, and install modern networked data communications. PhantomStrike is an active electronically scanned array radar developed for aircraft with tighter size, weight and power constraints than larger combat platforms.
Those additions are intended to allow the X-62 to test more demanding combinations of sensors, communications links and autonomous software. They could also enable researchers to evaluate how AI agents receive information from external platforms and coordinate their actions within a broader network.
Colonel Maryann Karlen, commandant of the US Air Force Test Pilot School, said, as quoted in the news update: “X-62 is continued proof that an agile, dedicated team can rapidly advance Department of War initiatives through intentional collaboration. Accomplishing these significant milestones is a direct reflection of the adaptability and warfighter focus personified by all members of the school.”
The US Air Force Test Pilot School and the Test Resource Management Center are due to hold the first X-62 Research Symposium at Edwards Air Force Base from 11 to 13 August. The event will bring together government, military, academic and industry personnel involved in future experimentation with the aircraft.
Not ChatGPT I suspect. Just a simple mechanism which delves into sensor data to provide a solution. Not that dissimilar to thousands of similar computer solutions before although these computers continue to get more powerful month by month.
People are waking up to the term AI which is really a rebranding of the term computer – which made everyone fall asleep.
Almost anything is branded as ai these days, just a buzz word for most of it than much different to what video has been doing for years which could present a false sense if security if we aren’t careful.
Would chat GPT be able to do this eventually. Therefore all lesser players would have higher end capabilities.
I think most people in the media and the industry have little grasp on AI and zero understanding that large language and visual models are not AI. If you show an algorithm enough picture of a dog or a mig 29 eventually it’s able to spot dogs or Mig 29’s. That’s all any of this technology is. No intelligence just brute force computing with enough data.
Five years ago everyone called it Big Data, The a few scam artists at open AI called it AI and that what the media runs with.
Mostly correct, but Big Data and LLMs are not really the same, in theme, form or purpose.
The term intelligent is also rather subjective; I mean there are plenty of humans that think lack have less brain capacity then a wedge of blue stilton.