U.S. Air Force Deploys Open Autonomy System Across YFQ-44A and YFQ-42 Combat Drones.

The U.S. Air Force is fielding its government-owned Autonomy Government Reference Architecture, or A-GRA, across early Collaborative Combat Aircraft platforms, including Anduril’s YFQ-44A and General Atomics’ YFQ-42. The move is designed to prevent vendor lock and allow mission autonomy software from firms like Shield AI and RTX Collins to be swapped, upgraded, and competed at software speed.

The U.S. Air Force has moved a key piece of its Collaborative Combat Aircraft (CCA) effort from concept to integration by confirming it is now fielding a government-owned Autonomy Government Reference Architecture (A-GRA) across multiple platforms, with Shield AI and RTX Collins positioned as the mission autonomy vendors driving early implementation. The service says the open architecture is already being exercised through semi-autonomous flight testing with General Atomics on the YFQ-42 platform and Anduril on the YFQ-44, a deliberate push to prove that mission software can be swapped and upgraded without being tied to a single airframe or prime contractor.
Follow Army Recognition on Google News at this link

Shield AI's Hivemind autonomy has been selected for the U.S. Air Force CCA program, supporting an open architecture approach across platforms like the YFQ-44, to deliver scalable combat mass in contested airspace (Picture source: U.S. Air Force).

The U.S. Air Force is trying to industrialize autonomy the same way it industrialized precision navigation and datalinks: by setting a common standard and then forcing competition above that standard. The service’s A-GRA effort is explicitly designed to prevent vendor lock by decoupling mission autonomy from vehicle hardware, so an autonomy “brain” can be fielded, replaced, or improved across different CCAs at software speed. In the Air Force’s framing, validating A-GRA across multiple partners is a cornerstone of its acquisition shift toward speed and a software-first mindset, where the best algorithms can be deployed rapidly on any compliant platform rather than being trapped inside proprietary architectures.

Shield AI’s role inside that construct is defined by its Hivemind autonomy software. In its February 13, 2026 press release, the company states it was selected after a competitive evaluation to support mission autonomy Technology Maturity and Risk Reduction efforts for CCA, and that Hivemind has already been integrated on Anduril’s Fury, designated YFQ-44A, to support system-level testing ahead of flight demonstrations expected in the coming months. That placement matters because Anduril’s Fury is positioned as one of the Air Force’s early CCA prototypes, so Hivemind is not being treated as a lab curiosity. It is being wired into an operationally relevant air vehicle and measured in the loop with real mission systems, real integration constraints, and the unforgiving timelines of flight test.

Shield AI describes Hivemind as core artificial intelligence software that assumes the role of a human pilot or operator, enabling unmanned systems to sense, decide, and act. The company draws a sharp line between classical autopilots and mission autonomy: instead of following a preplanned route, Hivemind is presented as capable of rerouting around no-fly zones, avoiding or engaging obstacles, responding to unexpected conditions, and completing missions without human intervention. Those claims point to an autonomy stack that must fuse onboard perception, dynamic planning, and real-time control under uncertainty, because the mission value of a CCA disappears the moment it needs perfect comms and a clean GPS picture. In practice, mission autonomy for a fighter-adjacent uncrewed aircraft is about managing time-sensitive tradeoffs: when to push forward, when to reposition for sensors, when to preserve survivability, and when to accept risk to protect the crewed force it is teamed with.

The Air Force’s open architecture push also clarifies how Shield AI is inserted into the broader program. Hivemind is described as A-GRA compliant and platform-agnostic, and Shield AI notes demonstrations of A-GRA-aligned autonomy across a mix of government and industry efforts, including work involving General Atomics’ MQ-20 Avenger, Northrop Grumman’s Talon IQ autonomous ecosystem, a U.S. Navy BQM-177 test aircraft, and even the Airbus UH-72A Lakota helicopter. Read through a CCA lens, this is less about marketing breadth and more about integration confidence: if autonomy can be made to interoperate across dissimilar airframes and mission systems, the Air Force gains leverage. It can compete autonomy vendors, pivot away from underperforming software, and scale the best behaviors across the fleet without buying a whole new aircraft.

Zooming out, CCA is the Air Force’s attempt to secure affordable mass and tactical options in a theater where peer air defenses, electronic attack, and long-range weapons punish predictable force packages. If a crewed fighter has to carry every sensor, every jammer, every decoy, and every missile, the force becomes expensive, finite, and easier to model. CCAs invert that problem by distributing payloads and roles across a team: some aircraft extend sensing, some complicate the adversary’s targeting picture, some carry weapons, and some serve as sacrificial scouts. The Air Force has been building toward this for years through human-machine teaming demonstrations, including a July 2025 event where pilots in an F-16C and F-15E each controlled two XQ-58A Valkyrie aircraft in an air combat training scenario at Eglin Air Force Base, explicitly to reduce pilot workload while expanding situational awareness and mission effectiveness.

The strategic advantage the United States is chasing is not merely more drones. It is decision advantage under stress. In a high-end fight, the side that can generate more tactical dilemmas per minute, while adapting faster than the adversary can re-plan, tends to seize the initiative. Autonomy and open architectures are enabling technologies for that advantage. A-GRA gives the Air Force a pathway to iterate tactics and behaviors across multiple vendors and platforms, while mission autonomy products like Hivemind aim to keep uncrewed aircraft effective when communications are degraded and the battlespace is changing faster than remote operators can manage. If the Air Force can field CCAs that are affordable, adaptable, and safe enough to trust around manned aircraft, it gains a force multiplier that stretches magazine depth, complicates enemy air defense planning, and reduces risk to aircrews without surrendering human command authority.

Written by Evan Lerouvillois, Defense Analyst.

Evan studied International Relations, and quickly specialized in defense and security. He is particularly interested in the influence of the defense sector on global geopolitics, and analyzes how technological innovations in defense, arms export contracts, and military strategies influence the international geopolitical scene.