Written by Tilak Chandika, a third-year law student at the National University of Study and Research in Law (NUSRL), Ranchi.

“Any pursuit of superintelligence has to be grounded in the core principle that if the AI we build is not helping humanity and not under human control, it’s not worth pursuing.” – Satya Nadella.

On September 13^th, 2026, the leaders in the AI race called for a conference on regulation of such technology, which is expanding in uncontrolled fashion and may lead series of events which deviate from the whole cause of its very invention. Anthropic CEO Dario Amodei emphasised that there are real dangers which are unfathomable and asserts that the industry has lied for too long abouts the risks of AI. The rapid militarization of artificial intelligence is outpacing the speed at which it is governed, across battlefields autonomous and semi-autonomous systems are being established to eliminate targets based on sensors and algorithms at speeds which are impossible for humans. Moreover, the AI race is pushing states to be better equipped with such technology for AI dominance, without such technology the arsenal at disposal is left to erode or has become dust.

Modern day warfare has transformed completely, to understand how militarization of AI works it is pertinent to analyse series of events and changes in policies that have taken shape. The United States of America has been a dominant global power in terms of military expenditure and research, in 2017, “Project Maven” was first initiated by the Deputy Secretary of Defence Bob Work under the working title of Algorithmic Warfare Cross-Functional Team. The core task of the team was to increase efficiency and reduce data overload within the Forces. The data included analysing tons of footages, data points from drone surveillance, pool of  intelligence tips and radar data etc. between 2020 and 2022, the project expanded beyond its basic functions to a comprehensive battle management system. Maven could then process massive streams of raw imagery, video, radar, and sensor data into actionable intelligence, now the time taken to detect, identify and strike a target has come down to weeks to days to hours to minutes and what officially is known as the “kill chain” is now rapid.

By 2024, the platform has achieved full operational deployment, it was directly integrated with field units and strike systems across U.S. Central Command, which now allowed forces to process live intelligence across departments and timelines to execute real time operations. Similar platforms took birth, such as “Gospel & Lavender” by Israel, “BAE Systems” by U.K. and “Ghost Shark & Loyal Wingman” by Australia etc.

The risks and restrictions of Militarization of AI are not fully known, yet, forces are excessively relying on AI systems. During war, personnel are under pressure to take the decision and with such information overload they are more likely to rely on AI systems, what is called as “Automation Bias”. This has been recorded and many events, one such incident occurred on February 28, 2026, the strike that hit a school in Iran, Minab, the strike killed 175 people (mostly kids) and these targets were misidentified as military targets due to outdated radar data that was fed to AI systems. Similar incident had occurred earlier this year when Israel attacked flagged Palestine residents, the AI system used by Israel i.e. Lavender had flagged more than 37,000 Palestinians as threats but when the attack took place the families of those targets were also killed in the operation. Such extensive reliance in AI was not the cause in itself because these systems are only as reliable as the data they are fed upon, but the real reason was automation bias and no regulatory mechanism to cross check the product of AI systems.

From a legal standpoint, the unregulated deployment of AI in military directly threatens the humanitarian law, the principles of the Law of Armed Conflict (LOAC) and additional principles laid down for the use of any new weapon under Geneve Convention are violated. Article 36 of the Geneve Convention specifically mandates for states to conduct formal legal reviews of any new weapon, means or method to warfare. These systems run in binary fashion, they can’t distinguish numbers and living people, nor are they trained for such purposes, it creates a challenge to trace back the decision and cause for such faulty actions that count as war crimes. Automated kill chains eliminate the possibility for diplomatic discussions, merely triggering the sensor and radar is enough for the system to take a decision.

Form legal jurisprudence standpoint, there are profound challenges when determining accountability, both legally and ethically, the existing frameworks and guidelines are not equipped to deal with such questions at stake. Moreover, the influence of defence contractors pose risk, as competition drives these private players and with unknown risks and restrictions it leaves space for potential abuse. The core issue boils down to absence of policy and governance at both national and international levels. However, at the main push towards the diplomatic efforts in recent times is the REAIM (Responsible AI in the Military Domain) summit. The efforts of this summit explore traceability, field-testing and operational explainability. The push for an internationally binding instrument to strictly regulate or prohibit autonomous weapon systems is move by the CCW (Convention on Certain Conventional Weapons).

The current set of discussions and policy frameworks focus on few foundational aspects that govern the use of lethal autonomous weapon systems, the primary mechanism that can deal with the existing threats are Meaningful Human Control Mechanisms, simply put, engaging personnel in the loop. Every decision, execution order or authority must be consciously taken by personnel who are constantly in the loop of events and data but not subject to automation bias. Training personnel is as important as regulating the systems, trained personnel are well aware of any systemic inefficiencies or if anything is seriously wrong with the kill chain. Strict adherence to Article 36 Additional Protocol I to the Geneve Conventions requiring formal legal assessment of any discovery of new weapon, means or method. Increasing the transparency of corporate institutions for full discovery and keeping track of new developments, as matter of concern for humanity. Framing foundational AI constitution, which specifies prohibited acts and means to such acts which are universally applicable. Establishing corporate accountability, defining guiding principles on Business and Human Rights. Profit must not become the powerful force behind wars; such AI race must never lead to abuse of existing dilemmas.