Artificial intelligence (AI) has emerged as the most transformative force in modern military innovation since the advent of the nuclear bomb. From predictive targeting, autonomous vehicles to automated command and control directives, artificial intelligence has embedded itself across the broad spectrum of military technologies. We see the likes of Unit 8200’s “Gospel” and “Lavender” systems be used to systematically identify and aid in eliminating suspected militants by making suggestions on areas to bomb, or the US Navy’s demonstration of naval dominance with unmanned vessels able to successfully launch targeted missiles. AI will be the centre of most future discussions relating to national security, surveillance and safekeeping, and thus warrants a discussion on the frameworks present for guiding responsible AI systems in the defense landscape of the European Union.
I. AI and The Global Balance of Power
While artificial intelligence is often seen as exclusively a technological breakthrough, its integration into defense signals a deeper shift in the global power dynamic.
The United States has a rich ecosystem of private innovation, and is the undisputed leader in foundation models and advanced chip technology. It is clear that the nation believes in the institutionalisation of AI being central to future economic and military prowess. That being said, the ecosystem is propelled by the hyperactive private sector with industry leaders (Google, Meta, OpenAI, Anthropic) driving impressive foundational research alongside critical expansion of necessary infrastructure such as standard interfaces, libraries, and toolkits for designing, training, and verifying AI algorithms. In 2023, the Department of Defense (DoD) dominated AI contracts with a whopping 657 contracts (up from 254), with other agencies being a distant second and third with 115 and 49 contracts respectively. From a potential value perspective, Brookings reports that defense rose from $269 million with 76% of all federal funding to $4.323 billion with 95%. The nation saw the 1500% increase in the DoD AI contract values which meant that the “DoD grew their AI investment to such a degree that all other agencies [became] a rounding error”. Following this line of inquiry, it is to no one’s surprise that as per Stanford’s HAI Centre, U.S. private AI investment hit $109.1 billion in 2024, a figure 12 times that of China’s $9.3 billion and 24 times the U.K.’s $4.5 billion.
The People's Republic of China has approached the AI arms race with its signature all-encompassive nation-first model that has the explicit goal to close the AI gap with the US by 2030. With its rapid rollout of high-performance data infrastructure, indigenous chip designs, unique frameworks like Baidu’s PaddlePaddle and overwhelming number of patents submitted, it is clear that applied AI is prioritised over blue-sky research. Leaders include Tencent, Alibaba, Baidu and MiniMax (currently targeting a $4 billion Hong Kong IPO). However, with Washington’s sharp restrictions on advanced semiconductors and cross-border initiatives with other sanctioned states like Russia, the nation is proving to be volatile grounds for pivotal startups like Manus fleeing to neutral states like Singapore. Regardless, the Republic has reached an estimated $98 billion in state AI investment, a reported 48% increase from the year previous as a result of integrating internet and telecom juggernauts to their national strategy.
Russia has not approached AI as a contest for global technological supremacy, but has primarily utilised it as a strategic lever meant to bolster its position in military and cyber warfare. It is largely because Russia has been “cut off from the global market” and is unable to compete on the same footing as China and the USA. Russia’s invasion of Ukraine meant the nation faced an onslaught from Meta, Microsoft, Amazon, Google, Starlink and startups like ClearviewAI that have collaborated to protect Ukraine from Kremlin-originated cyberattacks, migrate critical government data to the cloud, and help identify the faces of Russian soldiers. More recently, Russian intelligence did not foresee or prevent Operation Spiderweb, when the Ukrainians commandeered 117 drones into an airbase. Russia has similarly been employing the use of semi-autonomous weapons in the conflict to suppress this resistance, without all of the same investment and technological know-how received by the Ukrainians, with strong plans to utilise their funding and BRICS alliance to win this war with AI. As per the Ministry of Digital Development of Russia, Russia is starting to design a single trusted platform for data exchange and analysis. It is planned to allocate an estimated $13 billion alone to create a platform from the federal budget by 2030, with many other supportive initiatives for targeted AI development and collaboration with China.
Israel has long outperformed most nations in its warfare capabilities, and their foray into AI-enabled warfare displays the same focused approach to successful military conflict by the Israeli Ministry of Defense. IMOD launched a dedicated AI and Autonomy Administration in 2025 meant to support the IDF. IMOD reportedly developed AI tools meant for surveillance, facial-recognition and even bore the capacity to track targets of interest and triage their location from phone calls. Many hotly-debated scenarios of permissiveness surrounding autonomous weapons draw their ire from the events unfolding in Gaza. IMOD has an innovation hub linked to ex-Google, Meta, OpenAI members who reportedly aided the creation of such tools alongside numerous tech startups incubated in Israel. The launch of Israel’s national AI supercomputer in 2025 is an investment exceeding an estimated $133 million aiming to equalise AI innovation for startups and academics alike and will be led by AI cloud provider Nebius. Israel’s second phase National AI Program, launched in 2024, focuses on scaling R&D infrastructure, growing human capital, and embedding AI in critical public services and platforms further with permanence.
The European Union finds itself in a unique position where they cannot match the pace and scale of investment in AI seen in the United States and China, but they bear member states such as France and Germany who approach AI integration as a necessity and have the resources to act on it aggressively. The EU is a global leader in terms of its regulatory foresight, but is still dwarfed by the US in total private AI investment where in 2024 they received $19.42 billion and the US saw nearly 5 times that. Despite that, public investment continues as the EU launched the “AI Continent Action Plan” by committing an enormous 200 billion Euros over 5 years to develop unparalleled computing capacity and research. It is still an undeniable conclusion that such an ambitious initiative still trails the $109 billion private investment the US received in 2024 alone. Notably, France has emerged as the EU’s lead defense AI voice with a 2 billion Euro budget allocation to developing defense AI by 2030, swiftly establishing the AMIAD agency and with a 15 million research centre at the acclaimed Ècole Polytechnique. Mistral AI, the lauded AI company who can stand shoulder to shoulder with OpenAI and Google, signed memorandums with this agency and will help co-develop robotics and autonomous systems for defense.
II. Background: AI and Modern Military-tech Complex
Artificial intelligence has emerged as the most transformative force in modern military innovation since the advent of the nuclear bomb. From predictive targeting and autonomous aerial vehicles to algorithmic decision military decision making, artificial intelligence has embedded itself across the broad spectrum of military technologies. With operational activity in Ukraine, the conflict in the Middle East, and the South China Sea, this technology is no longer speculative for a decade down the line – it is today’s issue. Additionally, many military experts report the technologies garnering faster response times and reduced human casualties in specific types of missions (Etzioni & Etzioni). On the flip side, other field authorities have expressed concerns over black-box decision making (a lack of decision explainability), less human oversight, and outdated legal frameworks (Sullivan).
While tactical benefits – like faster response times and reduced soldier casualties – tend to dominate defense discussions, many AI safety researchers warn of much deeper, enduring risks. Among the most serious are catastrophic failure scenarios, which include autonomous misfires, loss of human oversight, and unintended escalation between militaries. These dangers stem from incongruity between the AI powering autonomous weapons and decision making platforms and the human leaders, in which the AI system may pursue unintended objectives in high stakes contexts (Hendrycks et al.). According to a 2023 study by Hendrycks et al., military AI misalignment could result in mass civilian harm, destroy geopolitical relations, and induce accidental war.
Operationally, artificial intelligence has sweeping applications. Most notably, many militaries utilize it for ISR – or Intelligence, Surveillance, and Reconnaissance – in which AI interprets imagery originating from satellites, drones, signals, and social media. Additionally, autonomous and semi-autonomous vehicles – like drones, UAVs, and ground robots – are increasingly being guided by real-time AI inputs. Further, both logistics and weapons integration have been focused on the military based AI revolution – whether through predictive supply-chain systems or strike recommendations (Rashid et al.).
As artificial intelligence becomes increasingly embedded in military decision making, unprecedented ethical and legal risks are introduced – many of which existing legal frameworks are ill-equipped to tackle. First of all, the black box nature of military AI systems has been known to produce outputs with unexplained origin, creating fervor across experts due to the possibility of a fatal mistake – and a lack of legal minutiae to determine accountability (Holland Michel). Additionally, reduced human oversight and blind trust among military officials has been reported to be an emerging problem in the landscape of war – effectively disregarding “human-in-the-loop” safeguards. Lastly, AI’s general use nature has created an ethical dilemma of dual-use – with experts noting blurred civilian-military boundaries, ethical risks, and export control loopholes (Black et al.).
Beyond tactical concerns, risks from AI misalignment must be taken seriously in defense contexts. Scholars like Hendrycks et al. warn of cascading failures across coordinated military systems – from autonomous misfires and goal drift to miscalculations causing escalation between nuclear powers. These risks are especially puissant when considering ISR, weapons guidance, and autonomous logistics platforms, where adversarial data, inadequate oversight, or model ambiguity can translate into fatalities and further warfare, ripping apart regions. Some key defense AI risks due to misalignment include goal drift, adversarial manipulation, autonomous escalation, ally misidentification, communication blackout fragility, and cascading effects (Hendrycks et al.).
III. The Role of Defense in the Civilian Sector: Palantir Technologies and Anduril Industries
Furthermore, the private sector has emerged as the leader and driver of the AI revolution, with companies like Anduril, AWS, and Meta taking charge – while governments contract them rather than innovate. While contentious if these changes are positive or negative, there has been a clear shift in power as a result of defense-tech innovation – or a lack thereof.
To understand the practical implications of AI adoption in the military, it is of paramount importance to examine the innovators at the forefront of this global revolution. Palantir Technologies and Anduril Industries – both of which are US based – offer distinct, yet complementary, views for how AI should be reshaping the defense industry. From the development of battlefield software to autonomous weapons systems, these ventures are far from mere vendors but strategic partners in shaping how conflicts are fought and deterrence is maintained.
Founded in 2003 to aid in counterterrorism and intelligence activities, Palantir Technologies is known for creating powerful data integration platforms. Some of its core technologies include Gotham and MetaConstellation. Gotham, one of Palantir’s flagship services, is not a weapon or a drone; on the contrary, it is a decision support system. It actively combines satellite imagery, drone feeds, signals intelligence, geospatial information, and even social media data into a single operating picture for military operators. Additionally, it provides AI assisted workflows for plan operations on every level with simulations of different courses of action to ensure data-driven decisions (Palantir). MetaConstellation, Palantir’s satellite focused AI platform, transforms a variety of satellite feeds into mission-ready intelligence through AI. Additionally, the technology is able to answer specific queries that are specific to various regions (Maçães). Together, these platforms indicate a profound shift in modern warfare activities – not through increasing firepower, but by streamlining intelligence and giving military professionals nearly superhuman situational awareness. Palantir’s deep integration into Western defense mechanisms is reflected not just in its innovation, but in its expansive economic footprint: the company’s stock price has more than doubled year-to-date to $151 – reaching a market cap of more than $350 billion. Furthermore, only two months ago, Palantir secured a defense contract valued at almost $800 million for Maven Smart System software licenses – with 5 other smaller contracts awarded in the last year (US Department of Defense). These figures highlight Palantir’s transformation from a Silicon Valley Startup to a central piece of the US military's technology infrastructure.
Anduril, which was founded in 2017, builds autonomous weapon and sensor systems. While the company offers a variety of state-of-the-art platforms, two of its most significant products are Lattice, an AI-powered battlefield operating system, and the Ghost drone, an autonomous aerial ISR platform. Lattice OS, the brain of Anduril’s operations, works to create a cohesive battlefield picture by integrating numerous mediums to support real-time decision making (Anduril). On the hardware side of things, the Ghost Drone is an autonomous quadcopter with onboard AI, allowing it to perform ISR missions without repeated operator input. The vehicle is capable of tracking moving targets, mapping terrain, and identifying threats – all autonomously (Anduril). Anduril’s scale and impact are evident in its rapid growth and contract success. In its latest funding round, Anduril raised in excess of $2.5 billion and was valued at $30.5 billion – more than double its valuation of $14 billion from 2024 when it raised $1.5 billion (Shetti). Additionally, the company secured a landmark counter-drone tech contract with the Marines in March, totalling over $640 million over 10 years (Wilkers). Though newer and smaller in scale than Palantir, Palmer Luckey’s Anduril is making waves in the defense tech industry, evidenced by mammoth contracts and growing valuations.
Palantir and Anduril represent two ends of the defense tech spectrum: with Palantir focusing on intelligence fusion and human-in-the-loop decision optimization and Anduril focusing on sensor autonomy and fast-action ISR. Strategically, these firms convey that effective military AI is not a singular solution – it requires software platforms, autonomous systems, and computationally quick decision making software. In the EU, there is currently no enterprise that is directly analogous to the very pertinent work that Anduril and Palantir do – at least not with the scale, dual-use adaptability, or public to private sector synergy of either firm. While major ethical and regulatory questions are raised about the production of this technology – especially with regard to dual-use – the EU must decide whether it will be a consumer of foreign systems or a creator of sovereign, ethically-focused innovation.
IV. Civilian Defense AI in Europe
While the United States had channeled AI investment into defense through bridging the public and private sector and venture-backed defense startups, Europe has largely opted for a civilian-first, research-oriented approach. For example, Horizon Europe has allocated in excess of €95 billion to research from 2021 to 2027. While its primary focus is on civilian R&D, which includes climate, health, and digital themes, some of its funding aims at supporting the development of AI, robotics, and dual-use technologies that could certainly be applied in a defense context. With funding being allocated in areas like autonomous systems, cybersecurity, and advanced sensing, it is clear that some projects funded by Horizon Europe have military utility under a civilian guise.
However, because Horizon Europe explicitly avoids directly funding military projects, most defense-aligned AI innovation is done through dual-use language or sectors adjacent to defense – like space or infrastructure resilience. On the other hand, the European Defense Fund – the EU’s primary source of militarily aligned R&D – has a significantly smaller budget over the same period of time. Although it represents EU-level defense cooperation – thus breaking up years of fragmented spending – the organization has funding of €7.3 billion for 2021-2027. While the EDF is a historic step towards a more centralized EU defense innovation system, it is hindered due to regulatory hurdles regarding defense tech and ethical concerns.
Europe’s regulatory framework, which includes the General Data Protection Regulation (GDPR) and the newly finalized AI Act, has positioned the EU as a global frontrunner in ethical AI governance. While certainly laudable from a civilian point of view, this legislation was not designed to govern defense use. For example, dual-use systems fall into high-risk categories, even when they are used to for the purposes of national security. Additionally, the strict compliance forced by these legal frameworks slow down rapid prototyping and defense deployment when compared to global superpowers, like the US or China. Furthermore, privacy rules enforced by the GDPR make real-time ISR harder, especially when they incorporate facial recognition or social media scraping.
According to the European Parliament’s 2021 policy study on military applications of AI, the EU currently has a paucity of strategic doctrine for defense AI; rather, it relies on fragmented national initiatives and dual use technologies under programs like Horizon Europe. Although Horizon explicitly avoids directly funding military projects, the report notes that civilian-focused research in areas like autonomous systems and surveillance often latent military utility (Franke). This indirect approach to military tech innovation reveals the caution and regulatory complexity in the EU, but also reflects a sense of structural weakness in Europe’s ability to compete with nations like the United States or China, who have a much more integrated R&D model.
While Europe has traditionally maintained a cautious stance concerning militarized innovation, an ecosystem of AI-driven defense tech startups is quietly emerging – most notably, Helsing and Safran.
Helsing, a German based venture founded in 2021, works to develop AI driven decision systems for defense. In its latest funding round – a Series D round led by Prima Materia – Helsing raised €600 million, increasing its valuation to in excess of €10 billion (Demetz). Supporting products like strike drones, autonomous underwater glover networks, and autonomous wingman systems, Helsing shows extreme promise in hardware development – with some having been deployed in Ukraine already. Specifically, the HX-2 AI Strike Drone has a range of up to 100km, a weight of 12kg, and a max speed of 220 km/h, all the while being able to carry anti-tank and anti-structure munitions and having full resistance to electronic warfare (Malayil). Additionally, the SG-1 Fathom Glider & Lura Platform – an autonomous underwater glover network powered by AI – can patrol up to 3 months submerged and is capable of detecting submarines and ship signatures that are up to 10 times quieter and 40 times faster than humans (Helsing).
Safran, formally established in 2005 and headquartered in Paris, France, is a leader in the aerospace, defense, and space industry – having a revenue of €27.3 billion in 2024 and in excess of 90,000 employees. The firm strategically acquired Preligens in 2024, thus integrating ISR analytics into Safran’s suite of services. Additionally, the company’s major products include the ACE – or Advanced Cognitive engine – which enables real-time target detention, classification, and tracking under challenging conditions, and AI.STAR, which is a sensor fusion platform capable of consolidating data from the air, land, and sea to support geo-referenced decision making (Safran).
Together, these firms demonstrate that while Europe lacks a Lockheed or Palantir-style defense juggernaut, it possesses the technical foundations and industrial base to nurture sovereign AI capability – so long as strategic support can parallel innovation.
V. Tensions in Defense AI Regulation and the Ethics of Warfare: Pushing for Further Guidance for Defense AI
As artificial intelligence becomes the centerpiece of modern warfare, European policymakers face a dilemma: how to balance strategic relevance with ethical protections and restraint. The EU leads globally in regulatory foresight – as seen with the AI Act and GDPR – but lags behind other global leaders in defense capability. Although it has fostered a tradition of ethical leadership in the EU, the general sentiment of civilian-first innovation may create vulnerability in a high stakes arms race. As a result, Europe risks becoming a consumer rather than a driver of military AI. Thus, the EU is in need of regulation that reconciles innovation and accountability. What is needed is a legislative mechanism that ensures human oversight, transparency, and lawful deployment while meeting the innovative and operational demands of defense AI.
To bridge the gap between ethical governance and strategic innovation in defense AI, the EU should consider adapting the Global Partnership on AI’s (GPAI) Security and Safety framework to the defense tech industry. Having been launched in 2020, the initiative garnered support from members like the EU, US, Canada, UK, Japan, and others. Specifically, the Security and Safety framework concentrates on high risk applications of AI which could lead to threats to societal resilience. Its core principles are grounded in human-centric values and prioritize international cooperation. Thus, it is a strong candidate for adaptation because it emphasizes values grounded in safety, rather than restriction. Additionally, the framework covers goal misalignment, adversarial attacks, and fail-safe mechanisms – all of which are directly related to the discussion on defense. Furthermore, it encourages auditable AI development – making it vital for accountability in high-stakes situations (GPAI).
Without proactive regulation, the rapid militarization of AI threatens to outpace existing ethical frameworks and legal oversight. Clearly, unchecked development can lead to the misuse of highly lethal autonomous weapons, unaccountable decision making in urgent situations, and normalization of lethal automation without a formal, democratic debate. The EU is uniquely positioned: it has a legacy of ethical tech regulation – evinced by the AI Act and GDPR – and has an international reputation as a leader in digital policy. Thus, as the EU shaped global norms on data privacy and digital rights, it must now fashion the moral boundaries of wartime autonomy – all the while not cutting off innovation and becoming dependent on allies. The future of defense tech should not be dictated by who can build the most efficient weapon, but by who can build the most enlightened system.
Work Cited
“Anduril.” Www.anduril.com, www.anduril.com/article/anduril-s-lattice-a-trusted-dual-use-commercial-and-military-platform-for-public-safety-security/.
“Anduril — Ghost.” Www.anduril.com, www.anduril.com/hardware/ghost-autonomous-suas/.
“Anduril Wins $642M Marines Counter-Drone Tech Contract - Washington Technology.” Washingtontechnology.com, 2025, www.washingtontechnology.com/contracts/2025/03/anduril-wins-642m-marines-counter-drone-tech-contract/403613/. Accessed 17 July 2025.
Black, James, et al. Strategic Competition in the Age of AI Emerging Risks and Opportunities from Military Use of Artificial Intelligence. 2024.
“Cluster 4: Digital, Industry and Space.” Research and Innovation, 2025, research-and-innovation.ec.europa.eu/funding/funding-opportunities/funding-programmes-and-open-calls/horizon-europe/cluster-4-digital-industry-and-space_en? Accessed 17 July 2025.
“Contracts.” U.S. Department of Defense, www.defense.gov/News/Contracts/Search/palantir/.
Demetz, Isabel. “German Defence Start-up Helsing Raises 600 Million Euros in Latest Investment Round.” Reuters, 17 June 2025, www.reuters.com/business/aerospace-defense/german-defence-start-up-helsing-raises-600-million-euros-latest-investment-round-2025-06-17/.
Etzioni, Amitai, and Oren Etzioni. “Pros and Cons of Autonomous Weapons Systems.” Military Review, Army University Press, June 2017, www.armyupress.army.mil/Journals/Military-Review/English-Edition-Archives/May-June-2017/Pros-and-Cons-of-Autonomous-Weapons-Systems/.
European Commission. “European Defence Fund (EDF) - Official Webpage of the European Commission. - European Commission.” Defence-Industry-Space.ec.europa.eu, 2025, defence-industry-space.ec.europa.eu/eu-defence-industry/european-defence-fund-edf-official-webpage-european-commission_en.
Franke, Ulrike. Artificial Intelligence Diplomacy Artificial Intelligence Governance as a New European Union External Policy Tool. June 2021.
“Gotham | Palantir.” Www.palantir.com, www.palantir.com/platforms/gotham/.
Hendrycks, Dan, et al. An Overview of Catastrophic AI Risks. 9 Oct. 2023.
Henshall, Will. “U.S. Military Spending on AI Surges.” TIME, 27 Mar. 2024, time.com/6961317/ai-artificial-intelligence-us-military-spending/.
Holland Michel, Arthur. “The Black Box, Unlocked: Predictability and Understandability in Military AI.” UNIDIR, 22 Sept. 2020, unidir.org/sites/default/files/2020-09/BlackBoxUnlocked.pdf, https://doi.org/10.37559/sectec/20/ai1.
Maçães, Bruno. “How Palantir Is Shaping the Future of Warfare.” TIME, Time, 10 July 2023, time.com/6293398/palantir-future-of-warfare-ukraine/?utm_source=chatgpt.com. Accessed 16 July 2025.
Malayil, Jijo. Interestingengineering.com, 2 Dec. 2024, interestingengineering.com/innovation/uk-new-hx-2-strike-drone. Accessed 17 July 2025.
“Palantir’s MetaConstellation Software Harnesses the Power of Satellite Technology.” Palantir, 2021, www.palantir.com/offerings/metaconstellation/.
Rashid, Adib Bin , et al. “Artificial Intelligence in the Military: An Overview of the Capabilities, Applications, and Challenges.” International Journal of Intelligent Systems, vol. 2023, no. 1, 6 Nov. 2023, pp. 1–31, https://doi.org/10.1155/2023/8676366.
“Safran AI Solutions Serving the Armed Forces.” Safran, 13 June 2025, www.safran-group.com/news/safran-met-ses-solutions-ia-au-service-forces-armees-2025-06-14. Accessed 17 July 2025.
“SG-1 + Lura – Helsing.” Helsing.ai, 2025, helsing.ai/lura.
Sullivan, Scott. “Targeting in the Black Box: The Need to Reprioritize AI Explainability - Lieber Institute West Point.” Lieber Institute West Point, 28 Aug. 2024, lieber.westpoint.edu/targeting-black-box-need-reprioritize-ai-explainability/.
The Global Partnership on Artificial Intelligence. “About - GPAI.” Gpai.ai, 2020, gpai.ai/about/.
Utkarsh Shetti. “Anduril Secures $30.5 Billion Valuation in Latest Fund Raise.” Reuters, 5 June 2025, www.reuters.com/business/anduril-secures-305-billion-valuation-latest-fund-raise-2025-06-05/.