France Plans AI 'Wingmen' Trials Next to Rafale Pilots by 2028

A French fighter pilot in 2028 may not fly alone. Two parallel tracks, cockpit AI on modified Mirage 2000D testbeds and a national request for information for a companion drone meant to fly alongside the Rafale. The unit of combat power is shifting from the single jet to the team one human commands.

France is not the first mover on this. The US Air Force's Collaborative Combat Aircraft (CCA) program, which pairs crewed fighters with autonomous wingmen, and the UK/Italy/Japan Global Combat Air Programme, which is building a sixth-generation crewed-and-uncrewed system, set the prior benchmarks. Paris is choosing a different shape: sovereign algorithms, a national drone, and a date that exists in law rather than on a flight line.

The two tracks converged this summer. In June 2026, the Direction générale de l'armement (DGA), France's defense procurement agency, issued a request for information for a multirole unmanned combat aircraft sized to fly alongside the Rafale: a 500-kilogram payload, a combat radius above 1,000 kilometers, and a top speed above Mach 0.8. In July 2026, the Ministry of Armed Forces launched HypAIRion, an experimental program flying AI and mission-system software on two modified Mirage 2000D testbeds. The 2024–2030 Military Programming Law, promulgated in August 2026, then named 2028 as the first experimentation year for Rafale "companion drones."

HypAIRion is the cockpit half of the plan. The Mirage 2000D airframes are testbeds for the algorithms that will eventually sit in the Rafale: target detection, target recognition, and decision support for the pilot. The drone half lives in the DGA request. A 500-kilogram drone with a 1,000-kilometer combat radius is, in shape, a cheap, expendable sensor-and-weapons platform that extends one pilot's reach into airspace too risky for a crewed jet. A 2028 first trial is the earliest moment those halves can meet, and even then it is an experiment rather than a contract or a first flight.

The procurement language is precise about that gap. The RFI asks industry for a multirole uncrewed aircraft; it does not buy one. The Military Programming Law names 2028 as the experimentation year; it does not schedule a first flight or a deployment. The Rafale F5 standard, the next planned Rafale block, with combat-drone integration, is targeted for the late 2030s. What 2028 will deliver is a documented first: AI in the cockpit on one track, a national companion-drone design on the other, and a calendar that names the team, not the airframe, as the unit of combat air power.

HypAIRion builds the pilot's AI; the DGA RFI shapes the drone. Neither track is, on its own, a teaming system. The work that has to land between now and 2028 is the data link, the autonomy stack, the rules of engagement, and the certification path that lets a French pilot command an autonomous aircraft under French law and NATO procedures. That integration work is also where the open question lives. AI target recognition in combat aircraft is a documented risk area: false positives, target discrimination under stress, and accountability for lethal action. Europe has not settled autonomous lethal-targeting policy across member states, and France is not asking it to. France is asking whether a sovereign AI stack, a sovereign drone, and a sovereign legal frame can produce a working team by 2028.

Industry has the RFI. The law has the date. The testbeds have the cockpit. The 2028 milestone will show whether the two tracks meet on a flight line, or stay two parallel programs with a shared calendar.