Saab Moves To Industrialize Combat AI After Gripen Trials

Saab has disclosed progress in flight testing its own artificial intelligence (AI) algorithms aboard the Gripen fighter, expanding on prior work with Helsing’s Centaur AI agent trials as part of its ongoing efforts to industrialize combat AI for the Gripen platform. The update, reported by Aviation Week, confirms that Saab is continuing its internal AI development program for armed operations on the Gripen and moving beyond lab and simulator work toward in-flight experimentation. No specific mission details, test conditions, or performance outcomes are provided in the Aviation Week piece.

AeroTime’s coverage adds new specifics to the in-flight activities, stating that Saab has flown AI agents developed entirely in-house on a Gripen E fighter, with at least one agent taking control of the aircraft during beyond-visual-range (BVR) combat scenarios. AeroTime notes that these in-air trials mark a shift from earlier tests that relied on an external partner, Helsing, through the Project Beyond program conducted in 2025. The article describes Saab’s own agents, the training environments, and the tools used to build them, and indicates that the latest tests involve Saab-developed agents rather than Centaur, though it does not disclose dates, sortie counts, or the exact nature of the BVR engagements beyond the general description.

According to AeroTime, the AI agents are produced by Saab’s Advanced Programs unit through what the company calls its AI Fighter Pilot Academy. The bulk of the initial training occurs in simulators where the agents engage in various air-combat scenarios and amass millions of simulated flight hours. Their decisions are continuously assessed against a defined set of objectives to refine responses to different situations, with the in-air tests representing a progression from simulated to real-world flight testing. Saab emphasizes that the Gripen E’s software architecture is central to this approach, enabling software updates and AI testing on a fighter already in service without the need for a separate experimental airframe.

AeroTime also highlights Saab’s broader strategy to apply AI in combat roles beyond simply replacing a human pilot. Saab positions its AI agents as capable of assisting crews in the cockpit or potentially controlling uncrewed aircraft that fly alongside manned fighters, suggesting a modular, multi-use approach to integrating AI in mixed-crew and collaborative operations. The article frames the initiative within Saab’s broader ecosystem, including the Swedish sovereign AI initiative Sferical AI, in which Saab is a founding partner, as a source of computing power and AI infrastructure to support training and development.

The two outlets together indicate that the Gripen E remains the testbed for Saab’s in-flight AI experiments, with the aircraft’s architecture enabling continuous software integration and testing of AI agents without additional hardware changes. Saab’s spokesperson or executives, represented in the reporting, emphasize that the objective is not to replace pilots but to augment cockpit workflows, situational awareness, and decision-support capabilities, as well as to explore the potential for AI to manage unmanned platforms that operate in concert with manned fighters.

Both sources describe the progression from earlier external AI testing to current in-house AI development and flight trials. The transition from Helsings’ Centaur-based testing in 2025 to Saab-developed agents during in-flight BVR scenarios marks a milestone in Saab’s attempt to industrialize combat AI—creating a self-contained capability that can be iterated rapidly within Saab’s own teams and flight programs, and potentially offering operators a pathway to enhanced mission execution and fleet-wide AI-enabled situational data processing.

While neither source provides a detailed account of the specific missions, flight conditions, the number of sorties, or precise outcomes of the AI-controlled flights, they collectively establish that Saab is actively moving AI agents from simulators to the cockpit of the Gripen E. The reporting underscores that this work is intended to be integrated with Gripen’s existing software architecture, ensuring compatibility with current platforms and enabling future expansions to broader AI-driven combat functions, including potential collaboration with unmanned systems in air engagements.

The reporting also clarifies that Saab’s AI efforts are conducted within its own Advanced Programs unit and are backed by internal training regimes and development environments designed to produce agents capable of operating in diverse air-combat contexts. The emphasis on millions of simulated flight hours as a prelude to real-world testing highlights the scale of Saab’s in-house training regime and the emphasis on rigorous evaluation against objective performance criteria before deploying AI agents in more complex or higher-risk scenarios.

Taken together, the articles portray Saab on a path to industrialize combat AI for the Gripen ecosystem, leveraging an in-house capability to train, validate, and fly AI agents in operational flight tests. The coverage indicates a broader industry trend toward AI-enabled combat systems and suggests Saab’s intent to keep Gripen E at the forefront of integrated manned-unmanned and AI-assisted operations, with ongoing work to refine how these agents complement human decision-making and potentially extend the reach of Swedish aerospace technology in future defense programs.