AI Wingmen Surge—But Who’s Accountable?

The Air Force’s move to field semi-autonomous “robot wingmen” is racing ahead, raising big questions about cost, control, and who really holds the kill switch in future wars.

Story Snapshot

The Air Force chose General Atomics and Anduril to build the first combat-ready Collaborative Combat Aircraft drones.
At least 150 “loyal wingman” drones are planned by decade’s end, with a goal of 1,000-plus over time.
Separate software vendors will compete to supply the artificial intelligence brains that fly and fight these aircraft.
Open-architecture autonomy is meant to prevent vendor lock-in, but real risks and accountability questions remain.

Trump-era Pentagon Bets Big on AI Wingmen

The United States Air Force has now moved from talk to action on drone wingmen, selecting General Atomics and Anduril to build the first operational Collaborative Combat Aircraft fleets.^[2] These drones are meant to fly alongside fighters like the F-35 and future F-47, carrying sensors and weapons while taking the biggest risks instead of a human pilot.^[4] Air Force leaders say the first three production lots should field at least 150 systems by the end of the decade, with long-term plans for around 1,000 aircraft or more.^[2]^[1]

Under President Trump’s second term, Pentagon planners describe this as a push for “affordable mass” in the air—many lower-cost drones instead of only a handful of very expensive jets.^[4] The idea is simple: China and other rivals are building up huge missile and air forces, so America needs numbers in the sky without breaking the budget. Each Collaborative Combat Aircraft is targeted to cost far less than a crewed fighter, bringing real firepower at a price point where combat losses are acceptable in wartime.^[6]^[24]

How The CCA Drones Work – And Why Two Vendors Won

After early concept work with five companies, the Air Force narrowed its Increment 1 competition to General Atomics and Anduril in 2024.^[4]^[6] General Atomics offers the YFQ‑42A, while Anduril’s design is the YFQ‑44A Fury, both officially designated as prototype fighter drones by the service in 2025.^[7] Within about 18 months of award, General Atomics flew the YFQ‑42A in August 2025 and Anduril followed with YFQ‑44A’s first flight in October, pushing the program rapidly into live testing.^[3]^[5]

Flight tests now focus on how these jets fly, how their autonomous behavior works, and how weapons and sensors integrate.^[3]^[16] The Trump Pentagon has gone further and awarded full engineering, manufacturing, and production contracts to both companies for Increment 1 airframes, four months ahead of schedule.^[2] That means these are no longer just experiments; they are the first wave of an operational fleet. At least one design will likely go into larger-scale production around 2027, but keeping two winners now also keeps competition alive on cost and performance.^[2]^[6]

The AI “Brain” Fight: Open Architecture or Hidden Lock-In?

What makes these drones different is not only the airframe but the autonomy—the artificial intelligence “pilot” that plans routes, avoids threats, and supports human wingmen. The Air Force is trying something unusual by separating the software from the aircraft. Anduril, Shield AI, and Collins Aerospace will now compete in timed phases to provide the core mission autonomy software that runs on both vendors’ drones.^[1]^[2] This “plug and play” approach aims to stop any single contractor from owning the whole stack.

Recent tests suggest this is more than a slogan. Anduril’s YFQ‑44A has already flown a mission where it completed tasks first under Shield AI’s Hivemind autonomy and then switched mid-flight to Anduril’s Lattice system, repeating the mission profile before returning to base.^[11] That is direct proof that autonomy packages can be swapped without rebuilding the jet. For conservatives worried about decades of Pentagon vendor lock-in, this modular design—part of a broader open-systems push sometimes called a modular open systems approach—offers a rare sign that the bureaucracy is at least trying to keep leverage over big defense firms.^[5]

Opportunities and Risks for Warfighters and Taxpayers

For the warfighter, loyal wingman drones promise more shooters, more sensors, and more chances to survive a fight with a near-peer enemy. Air Force planners see these aircraft as key to maintaining air superiority against China while keeping American pilots farther from the most dangerous threats.^[4]^[7] Market analysts also note that military drone spending is exploding worldwide, with autonomous, “attritable” systems becoming central to future force structures because they can be lost in combat without collapsing the budget.^[24]

The U.S. Air Force awarded production contracts to General Atomics and Anduril to build its first fleet ‌of semi-autonomous Collaborative Combat Aircraft (CCA), moving a program that began just over two years ago from prototype to full-scale manufacturing.

USAF awarded…

— Aeronews (@AeronewsGlobal) June 18, 2026

There are still serious concerns. Many defense programs show great prototypes, then stumble on cost, reliability, and sustainment once real production begins.^[21]^[27] Autonomy also raises deep accountability issues: contractors write and help maintain these complex systems, blurring lines of responsibility if something goes wrong in combat.^[22] For constitutional conservatives, that means oversight matters. Congress will need to watch both spending and rules of engagement closely to ensure these AI wingmen serve American interests, protect our troops, and never replace human judgment where life-and-death decisions are made.