Defence AI: The Challenge of Sovereignty

Organization : Ministère des Armées / Published : November 14, 2024

Revolutionary like the atom in its time, artificial intelligence (AI) is making a growing place for itself on the battlefield. Already at work in the Russo-Ukrainian conflict and within the French armed forces, it is reshaping the art of war and should impose itself as an indispensable ally in the face of the massive flow of data that characterises our era. Here are some explanations.

Dans les champs immatériels, l’IA s’impose comme une arme d’influence et de désinformation. - © Olivier Berg/DPA/AFP

The Russo-Ukrainian conflict has become the "laboratory" of a new kind of war. A war where the most powerful army is no longer necessarily the one with the greatest strength and equipment, but it could be the one capable of analysing and exploiting the largest amount of data in real-time. The Ukrainian government, through its Minister of Digital Transformation, Mykhailo Fedorov claims: "Today we are a kind of training ground for the use of artificial intelligence." On the Ukrainian front line and in the decision-making loops, AI has quickly imposed itself as a bulwark against the mass of Russian soldiers and equipment. While it does not replace traditional warfare, this technology is now an indispensable ally in several areas: decision support, autonomy of weapon systems, observation, logistics, supply, cybersecurity, personnel health, human resources, to name but few.

Un soldat ukrainien utilise un drone Vampire, dans l’Est de l’Ukraine, le 2 février 2024. © Dmytro Smolienko/NurPhoto/AFP

Un soldat ukrainien utilise un drone Vampire, dans l’Est de l’Ukraine, le 2 février 2024.

While battlefields are not yet populated with Terminator-style killer robots, artificial intelligence algorithms are making a growing place for themselves in weapon systems. Due to the integration of digital technology, the military now finds itself managing an unprecedented amount of data. This is due to the proliferation of sensors, i.e. any device that allows the collection of information (the fuel of AI). These can be drones, satellites, mobile phones, radars, or cameras embedded in vehicles. These devices, mostly democratized and widespread globally, generate enormous volumes of real-time information that humans cannot process and analyse quickly and efficiently. Only AI currently has the computing and analytical capabilities to exploit ever-increasing amounts of data (inter-service, inter-allied, diplomatic, industrial) and meet the tactical needs of immediacy and precision. In this context, Admiral Pierre Vandier explains that AI will serve two purposes in the armed forces: "To save humans time on laborious tasks and to help with decision-making."

AI in Drones

The massive use of drones illustrates this new paradigm. For more than two years, they have been proliferating by the tens of thousands on the Ukrainian battlefield. On the lookout for the slightest movements of Russian troops, the drone is particularly effective for the positional warfare which the belligerents have been waging since the spring of 2022. Used by both sides, these low-detectable aircraft capable of flying up to 1,000 km from the operator constantly monitor enemy positions and collect large amounts of heterogeneous information that is processed in the rear thanks to artificial intelligence. This information sheds light on the battlefield, giving it an unprecedented transparency, dispelling what the Prussian military theorist Carl von Clausewitz called the "fog of war", which refers to the uncertainty surrounding combat. For Admiral Pierre Vandier, "the new fog of war lies in the adversary's ability to exploit the data gathered on the battlefield. The crux of the matter is the quality and speed of decision-making."

Since the early stages of the conflict, this strategic use of data has been quickly understood by the Ukrainian military which called on companies in the new technology sector to help them. Elon Musk, through his constellation of Starlink[1] satellites, provided the Ukrainians with a fast and secure internet connection on the front line, allowing them to counter the power and network cuts caused by the Russian military. This service has become a major support for the Observe-Orient-Decide-Act (OODA) loop, which allows coordination between command and control authorities, information flows, and the decision-making process using digital tools during the battle. "Without Starlink, we would have lost the war," said a Ukrainian military commander in July 2022. At the same time, Amazon secured the Ukrainian state's administrative and economic data by downloading 10 million gigabytes of data to the cloud[2].

Once the internet connection had been established and sensitive data were secured, the Ukrainian government counterattacked. It made an application available allowing military and civilians to upload photographs, videos, and any other information concerning the movements of the Russian armed forces. Coupled with satellite images and others from drones, hacking, or open sources like social media, this data is used for targeting operations with the help of AI-integrated software. This AI can even generate strategic and tactical recommendations, thus accelerating the OODA loop.

Beyond intelligence, AI-integrated drones are formidable weapons. Equipped with a grenade or a rocket warhead, they are capable of causing heavy damage to the adversary. Some can also bypass Russian jamming capabilities and operate without the operator if necessary, to continue the mission in a contested electromagnetic environment. In terms of equipment, the needs are considerable. Some Ukrainian authorities mention a need from 100,000 to 120,000 drones per month. According to the British think-tank Rusi (Royal United Services Institute), specialising in security and defence, it is estimated that Kiev loses nearly 10,000 drones a month. The reason is a frantic race for armament and information to keep the upper hand.

In this field, the belligerents have been fighting fire with fire. The lifespan of drone software on the Ukrainian front is less than 12 weeks before identification and the possible exploitation of a flaw by the adversary. "Keeping up the pace requires technicians and engineers to ramp up their creativity," explains Commander Vincent Sébastien, deputy director of the Ministry of Defence's AI Agency (Amiad). "Ukrainians are good mathematicians and programmers capable of tinkering with AI in a garage with very few resources. This relationship with technology, less top-down than in the Russian military, allows them to experiment and quickly test new strategies," he continues.

[1] Satellite internet access provider designed by the American aerospace manufacturer SpaceX.

[2] Remote storage solution.

A Weapon of Disinformation

Far from the fighting, in immaterial fields, AI also allows the falsification and manipulation of videos, photographs, or audio files. These contents, disseminated on social media or the internet, are used by Russia or other malicious players for propaganda and disinformation purposes. Deepfakes[1], for example, are difficult for the human mind to detect. The beginning of the war in Ukraine was marked by the dissemination of an AI imitation of President Volodymyr Zelensky, in which he called for laying down arms, which had been injected during a Russian cyberattack on Ukrainian communication stations. For Emmanuel Chiva, the Chief of Defence Procurement, these methods illustrate "the generalisation of AI as a weapon of disinformation and influence. This war teaches us that we must equip ourselves with the means to detect and thwart disinformation," he says. This issue is shared by many democracies around the world, regularly targeted by smear and manipulation campaigns on social media. In response, AI-based software, once again, will be able to study these contents (themselves generated by AI) and establish their truthfulness.

[1]For video infox, which is a video or audio recording made, or even modified, using AI.

The Challenge of Sovereignty

Like Ukraine, the French armed forces have been working on artificial intelligence for many years. As early as 2018, the French Ministry of Defence committed investment and launched nearly 400 projects in this area. The acceleration of the maturity of this technology now forces France to move to an industrial scale, so that the use of AI spreads in administrative tasks and operational missions.

To achieve this, Defence Minister Sébastien Lecornu presented the ministry's strategy in this area on 8 March 2024, at the École Polytechnique site in Palaiseau (Essonne). "The technological leap represented by artificial intelligence is undoubtedly the one that will revolutionise the way we wage war. Or even, more importantly, to avoid it, like the atom in its time," he recalled.

On the same day, he announced the creation of Amiad. Its mission is to enable France to sovereignly master these technologies so as not to depend on other powers, as General de Gaulle wanted in 1945 when launching our deterrence policy. This agency, based on the École Polytechnique campus in Palaiseau for its "research" component and in Bruz (Ille-et-Vilaine) for the "production" component, should ultimately have 300 people, including researchers and military personnel. They will be able to rely on "the most powerful supercomputer dedicated to AI in Europe," according to the minister. Installed at Mont-Valérien, in Suresnes, it will allow "sovereign processing of 'secret' data," but will also benefit other ministries, as well as companies in the defence industrial and technological base. The armed forces will also be able to test AI embedded in weapon systems "while maintaining a high degree of protection," the minister says in detail, while defence industries will also be able to use this supercomputer and feed it with their data, "without fear of espionage." Some defence companies, both large and small, will thus be able to "work in a secret militarised space."

As to the new director of Amiad, Bertrand Rondepierre intends to professionalise the use of AI in the armed forces. "Today, AI is no longer just a matter of innovation, but must be considered as a technology for operational applications. Also, Amiad's mission is not to push a vision, to convince or evangelise the Ministry on AI issues, but to contribute to the transformation of the armed forces. In short, it is about equipping ourselves to win the war."

A wide range of uses

Building on this momentum, use cases are multiplying within the French armed forces. The French Navy, for example, uses AI to help acoustic warfare analysts, the "golden ears," to sort sounds out. The objective is to only focus their attention on the useful, high-value signals, on which they can bring their expertise. Initial feedback shows a time saving of nearly 98%. For Commander Vincent Magnan, director of the Acoustic Interpretation and Recognition Centre (Cira), the use of AI in the French Navy is due to two major contextual elements. "Firstly, maritime traffic is increasing significantly. Secondly, acoustic detection equipment is more and more numerous and efficient. These two factors considerably increase the amount of data to be processed and become a major challenge." The figures illustrate this change in scale. In 2020, Cira collected 1 terabyte (TB) of data per year, compared to 10 TB in 2024-2025 and potentially 100 TB by 2040.

L'IA assiste « les oreilles d’or » de la Marine nationale dans l'analyse des détections acoustiques. © MT Jérôme Guégan/Marine nationale/Défense

L'IA assiste « les oreilles d’or » de la Marine nationale dans l'analyse des détections acoustiques.

Another area of application is collaborative infovalorised combat. The future of air-land combat, it will pass on the information in real-time between all units deployed on the ground. The goal is to transform the armed forces into networks of interconnected systems, capable of quickly adapting to a constantly evolving operational environment. The Scorpion[1] programme for the Army and Scaf[2] for the Air and Space Force are perfect illustrations of this. In these vast systems, AI will be in charge of collecting and merging data from multiple sensors (radars, cameras, vehicles, etc.) to provide a complete real-time picture of the operational environment. At the same time, AI will help the military make informed decisions by proposing scenarios and tactical options based on the situation.

[1]For Synergy of contact reinforced by versatility and infovalorisation.

[2]For Future Air Combat System.

Représentation du système du combat aérien du futur.  © Direction générale de l’armement

Représentation du système du combat aérien du futur. 

"Human Responsibility"

In both cases, drones (aerial or terrestrial) will play a key role and will have great autonomy, even if engagement remains the prerogative of humans. "AI does not have the right to define or modify the mission it has been assigned. This is called the permanence of command responsibility," explains Emmanuel Chiva. To mitigate fears of replacing humans with machines, France established a defence ethics committee in 2020, whose one of the first opinions focused precisely on lethal autonomous weapon systems (Salia). The document, dated 29 April 2021, emphasises that an autonomous machine does not set its own rules: "The highest values of our civilisation, such as our constitutional order, imply that human responsibility is engaged in all circumstances."

Kévin Savornin


Featured