Artificial Intelligence (AI) and International Humanitarian Law (IHL)

Introduction

The history of humankind is characterized by violence and conflict. Why? They also served as catalysts that brought about changes in society over time. The source of international humanitarian law (IHL) is the key point of emphasis in this blog post. The adoption of a legal framework that predominantly comprises the four Geneva Conventions of 1949 and their associated protocols, signifying the explicit recognition that in times of military necessity, human rights and Armageddon, conciliation must be maintained. Even in times of war, actions are subject to certain limitations. We are continuously evolving as a species, with technological advancements being the primary driver. It is no wonder that technology has influenced the nature of warfare. Considering the impact of new technologies, particularly AI and artificial intelligence (AI), is crucial to understanding how they affect the almost universal IHL framework for armed conflict situations.¹ Due to the seemingly limitless technological possibilities, it is uncertain if there are any limitations in using AI for combat. This worries to be asked how AI is interfering with current IHL. If yes, what are the consequences?

Definition and foundations of artificial intelligence

Human cognition, planning, or reasoning are required to perform tasks that would typically only be accomplished by computer systems through artificial intelligence. ChatGPT is a famous example of an AI system, but there are many other examples.

The foundation of AI is algorithms. An algorithm is a set of rules that impose instructions or behoove arbitrary logic to give resolving an issue or answer to ambiguity.

A machine learning system can generate its own instructions by utilizing data that it has been trained on. Using these instructions, the system solves problems that it is currently solving. There are various ways in which the software writes itself. Machine learning has become a new area of focus in AI technology. Certain machine learning systems acquire knowledge during their usage for a specific task by taking into account the environment they are operating in.²

Machine learning’s underlying nature means that the AI system may not always react identically to the same input, unlike simple rule-based algorithms. The system will be unpredictable.

The ‘black box’ of machine learning systems is a significant problem that needs to be addressed. It can be challenging to rationalize a system’s output in retrospect, even with known inputs.

Ai strategies and applications in armed conflict

AI is a major investment for the military, and it has already been utilized in warfare to provide intelligence or as ancillary elements in weapon systems.

In three areas where the ICRC has identified AI being used by armed actors in warfare, humanitarian questions are of significance:

The incorporation into weapon systems, particularly self-sufficient weapons, is essential. Application in cyber and information warfare.

Underpinning military ‘decision support systems

AI has been most extensively utilized for military purposes, particularly in the context of autonomous weapon systems. There are worries that AI could be utilized to launch an attack on someone or a vehicle. For example.

The ICRC has called on governments to implement new international regulations that would prohibit the use of certain autonomous weapons and limit their use, including those controlled by AI. Additionally, the risks associated with the application of AI in cyber and information warfare, as well as decision-support systems are being less discussed.

Human-centered approaches to AI usage in armed conflict are necessary for the international community, as it may harm civilians. Why?

Contributions of ai to military operations

The use of AI-powered software in creating analysis to aid military decision-making is possible through a computerized decision support system.

These systems collect, analyze and combine data sources to identify individuals or objects, assess behavior patterns, provide recommendations for military operations, or predict future actions or situations.

By analyzing drone footage and other intelligence, an AI image recognition system could be utilized to identify military targets.

These AI systems can be utilized to determine the target attack type and timing. The use of AI based systems in military decision-making regarding nuclear weapons has sparked alarming claims.³

There are those who assert that decision support systems can assist human decision-making in a way that promotes compliance with international humanitarian law and reduces civilian risks.

Current state of ai integration in military capabilities

People tend to neglect the practical applications of emerging technologies and instead focus on their present capabilities. As an example, one can go along with far-reaching debates about the full automation of warfare, where robots and unmanned weaponry make human involvement completely unnecessary. This gloomy scenario could be possible in the future, but it is not currently happening and humans are still dominant. However, The question at hand is not about human participation, but rather about the extent of it. This is important for now.

Even though there have been land and sea wars for centuries, AI-related issues are still in their early stages of development. Nevertheless, numerous practical AI applications in military operations are already established, such as the recent statements of Israeli army officials regarding the integral role of artificial intelligence in their military capabilities. Additionally, There are numerous applications of military AI in the modern world.? From aiding with the targeting cycle and hostilities, cyberspace attacks, intelligence surveillance, reconnaissance, and deployment of AI in ‘information warfare.’ They span various roles. To illustrate the latter in reality, one can see the use of fake videos to manipulate or deceive the public during the Russia-Ukraine conflict.

The international humanitarian law perspective

The normative framework of International Humanitarian Law is founded on basic principles, which have significant implications for its credibility and ability to influence state practice. The following principles can be summarized in a didactic manner:

(1) Humanity: respect for human dignity and safeguard against acts of violence or intimidation;

(2) Distinction: ensure that the combatants are clearly and objectively distinguishable from civilians;

(3) Force must be proportional to the military objective; such actions should not cause unnecessary harm or suffering to those who are not directly involved in the conflict; and, finally, exercise should be precautionary.

The Fourth Hague Convention of 1907 included the Martens Clause, which was originally drafted by Russian delegate Friedrich Martriens and published at the First Peace Conference in The Haguer in 1899.⁴

The first observation about this field of international law may seem natural, but this applies very well to the matter, that is, people are at the center of IHL. There is a perennial discussion among the international people how to find a reliable ways to cover military requirement and human ideas in accordance with the goals of IHL, which is to regulate the conduct of war and states during the war, it is not banned. Machines, according to the definition, are not humans, and their existence cannot question the integrity of the long -established judicial and ideological principle of IHL, who are laboriously forged in favor of individuals, with a view to gaining hope of safety in armed struggles. And if people are the main objective of this security, then the responsibility of fulfilling legal obligations from the states should not be transferred to machines or inanimate creatures. The IHL’s principles will require the minimum degree of human control and supervision, so that to open the possibility of intervention after activation of the system to be employed in a particular military context. The additional protocol for 1949 Geneva conferences on the protection of victims of international armed conflicts provides especially in Article 35 for the original norms of IHL, stating that in any armed conflict, it is not “unlimited” to struggle to choose the rights or methods or means of war in any armed struggle. The use of weapons, projectiles, materials and comparison methods that can lead to “luxurious injury or unnecessary pain” is prohibited. Similarly, it is a ban on the use of war methods or means, which are intended, or may be required, “broad, long -term and severe damage to Natural involvement.

In addition, in relation to the Protocol of Geneva I, Article 36⁵ directly addresses the issue of new weapons, whenever a highly contracted party provides for any research, study, development, acquisition, or “a new weapon, means, or the method of war” for the need for a legal review to adopt. In such a case, the state will have the responsibility of determining whether its employment will be banned by protocol or by any other rule of international law applicable to that state. It should be told that Article 36 does not represent universal panacea for the case of autonomous weapons. Some boundaries associated with this section should be remembered. Its application is a measure of national implementation, which basically means that it depends on the states under this legal responsibility to determine compatibility with the international law of new weapons, methods and means of war, whether there is a multilateral interaction regulation on the matter or not. Nevertheless, so far there is very little support from the states. Today only six countries (all those western states) have announced the adoption of processes to implement Australia, Belgium, Netherlands, Norway, Sweden and the United States. Germany and the United Kingdom claim to do so, but they did not disclose the internal mechanisms applied at that end. The confidentiality of the results of the legal review done in this way does not give adequate assurance to other states, which have no means that they verify to verify their disposal and clearly confirm how and what is actually investigated.

Article 36 does not consider comprehensive moral issues that are associated with the entire debate about human-masine interactions.⁶

After saying this, many authors (many of them are quoted in this letter) believe that according to IHL it is possible to raise serious objections on autonomous weapons. The main ideas arising from this important perspective can be more clearly seen if we return to the basic principles mentioned above.

Objection to autonomous weapons according to humanity:

Humanity: Machines according to the definition exclude human elements from the very beginning and for this reason they are in a different category.

Discrimination: Machines will probably not be able to separate fighters from citizens in all situations.

Proportional: Machines will hardly understand the context with proper confidence to correctly evaluate the overall picture.

Need: Machines cannot decide themselves whether it is necessary or not, as this assessment usually means a political decision that can be on the ground. • Precaution: Machines will have to program to decide in real -life situations, but there is no certainty whether it is possible or not.

In addition, in addition to IHL, the International Human Rights Act is extremely important in debate on autonomous weapons. Not only because the issue gained momentum with a session of the Human Rights Council in 2013, but the top of the centrality of the human person’s rights between the fundamental rights of the person. Machines, equipment, computers and robots are not moral agents nor do they have compassion, sympathy or intuition, the qualities contained in humans. An algorithm alone should not decide about life and death (Bhuta, N. Et al., 2016). Robots can help people live a full, respectable, dignified and healthy life. Our duty is to ensure that these benefits are transmitted properly, whenever possible, preventing abuses, adopting rules if necessary, and determining moral boundaries that require public conscience.⁷

International discussions under the convention on certain conventional weapons (ccw)

Against this background, our investigation on autonomous weapons should take into account some basic parameters. First of all, to accept that the automation used in military applications is already a reality, as fully displays the examples above. Rarely a specialist is ready to suggest the possibility of a spontaneous reversal of this process, although it is necessary to clarify from the beginning that automation is not necessarily complete autonomy of the weapon system. Second, to assume that autonomous weapons may have non-conscious volatile effects, when they are deployed in the battlefields, including strategic instability, landscape of unexpectedness, and a lack of control without human intervention to neutralize them in case of fundamental changes of high political decisions or conditions. Considering the current pattern of international conflict, we are looking at more and more tragic house wars, complex asymmetric struggles, violent urban battles and various types of contradictions, it may be a good case that the future deployment is the most likely to be the most likely to be affected by future deployment in developing countries and poor areas.

And third, to identify the fact that the states have a universal obligation to honour IHL. This obligation is not done by any scientific or technological development and will not be cancelled, although it can be innovative and revolutionary. It should be reminiscent of the convenience of an independent interaction, accepted, and agreed multilateral framework, which is valid for all states to address this issue in a broader way. Currently, the International Forum has been chosen by the states that the prohibition on the use of some traditional weapons or restrictions on the use of some traditional weapons to discuss the problem, which may be considered highly harmful or may have indiscriminate effects. It is usually briefly known, the United Nations Convention on Some traditional weapons (CCW) were opened in 1981, implemented in 1983 and currently 125 highly contracted parties (Brazilian is a party since 1995). Visualized as a framework convention, CCW is complemented by additional protocols on traditional weapons specific categories. This flexibility allows more space for adaptation in front of technological development, which can lead to the rise of new weapons, ammunition and related military products.⁸

Five additional protocols for CCW were adopted by signature states to deal with the following types of weapons specifically:

Protocol I: non-addressable pieces (1980);

Protocol II: Prohibition or restriction on the use of mines, booby-trap and other equipment (1980);

Protocol III: Fire -fire weapons (1980);

Protocol IV: blind laser weapon (1995); And

Protocol V: Explosive remains of war (2003). The informal discussion on autonomous weapons under CCW began in Geneva in 2014. For the first time in 2017, a group of government experts resolved the issue under the chairmanship of India’s Ambassador (Catton, 2015, P.41) to resolve the issue. The decision should be taken unanimously and so far there has been very little progress. There is a disagreement in the form of expected definitions, functioning and real boundaries for talks in terms of coverage and scope between other “red lines”.

There is no normal approach to what a restriction or prohibition should be imposed on autonomous weapons to any restriction or prohibition. The most advanced military powers are cautious about the hypothesis of starting serious restrictions on the use of these technologies (Russell, 2016) (at least to say). Among the five permanent members of the United Nations Security Council, only China has indicated openness to interact on a new protocol on autonomous weapons, although the details of the proposal and the objectives behind the Chinese status are also to be better evaluated (Kania, 2018). One of the advantages of this platform for debate after all is exactly the discovery of areas of convergence, even though at this stage they do not seem enough to underline an agreement in the short term. For example, there is a consensus that autonomous weapons should follow the international law of armed conflict. Most countries also advocate the need for meaningful human control in important tasks, especially in terms of selecting and attacking goals (ICRC, 2016, Horovitz and Shere, 2015). The challenge lies in demanding clarity if the status quo can bring a proper solution to maintain, without any major international regulation, or if there is a chance to succeed with options or lump sum restrictions on the development of these weapons. The political and technical feasibility of any of these proposals remains to be seen (Stop Killer Robot, 2018).⁹

Some people say that an adjournment would be better to immediately ban as it can give a “stagnation”, while the states assess what technology will look and in which direction the directions can be seen in the coming years. But, even, who can ensure this and verify that a sub -institution will prevent research and development for military purposes by actors wishing to use privacy for its benefit? The recommendations of the 2017 report of a group of government experts bring some incentives, due to the creation of most problems, although a clear roadmap for international talks is still missing. The CCW, reported in the report, “is a suitable structure to deal with the issue, which may indicate that the current efforts will continue for time with these lines. Also positively recommends that IHL remains “fully implemented to all arms systems” and the responsibility for the deployment of any weapon system in the armed conflict remains with the states, which should ensure proper accountability for fatal action as per the international law, especially in IHL. The report also highlighted the fact that, accepting their double -nature and rapid growth, the civil use of these technologies should not be a hindrance to progress. Discussion on the characterization of these systems should promote a “general understanding” on their relevant concepts and characteristics and in the end, possible options to address the human and international security challenges generated by emerging technologies in the field of deadly autonomous arms systems, at this time, “without prejuding the policy results and moving forward to take forward, the present, and future proposals.¹⁰

Brazil has participated in a group of CCW government experts through its diplomatic mission in Geneva, with the presence of military briefcase of the Ministry of Defense. The position of Brazil suggests that autonomous weapons should have “meaningful human control” and, until this end, legally binding equipment interactions should be started as soon as possible, which should be started without excluding the intermediate proposals or initiatives already. Important mass on this subject in Brazil, however, is far from the ideal if we consider the increasing importance of discussion on artificial intelligence, peace and security. It is one of the top concerns for the future of war nowadays, but it has recently begun to attract the attention of Brazilian writers, who do not belong to the technical location of AI experts, among the university professors, independent analysts, students, and international laws, strategy and diplomacy researchers. It should be noted that the National Commission for the spread and implementation of the International Humanitarian Law in Brazil, was created by decree in 2003, but still rarely known to the general public, the competent authorities are responsible for the measures required to disseminate and implement IHL in Brazil, especially for the Geneva Convention of 1949, 1949, the Geneva Convention, Additional Protocol and III. Seven Ministry is a member of the National Commission (Foreign Affairs, Justice, Defense, Health, Education, Culture, and Human Rights), as well as head of the President of Republic (CASA Civil).

Representatives of the Congress (Federal Senate and Chamber of Deputy), Red Cross (ICRC) and Brazil’s Red Cross also sit in the meetings of the National Commission. In view of the new issues being addressed by IHL, technological innovations also include immediate need to adapt the existing standards to deal with the means of war and methods in the methods of war, a subcommittee on new war technologies was created within the National Commission in 2018. Its main objective is to raise awareness and gather information that can contribute to the debate around the application and analogy of such innovations with international law, especially IHL. It is believed that a major engagement of various stakeholders of the government, academics, private sector and civil society can help further promote interest in research programs focused on strategic, moral and legal aspects among other new technologies, thus contributing to strengthening the posts of Brazil in international organizations and multi -level institutions.

Future scenarios and potential risks

We all risk are affected in any way if inactivity is handled to such a sensitive issue of international security. The immediate risk will be that, sooner or later, if not regulated in any satisfactory manner, completely autonomous weapons will eventually be developed and will become rapidly sophisticated. Equipped with advanced AI, smart weapons are believed to strictly follow their programming, it can be a problem due to flexibility or lack of human decision when facing changing situations. If the operator will be able to remember them, for example, the commander wishes to end an attack for some reason? The dizziness of a machine can cause an unexpected increase or conflict in the case of false alarms, mistakes or accidents. Canceling of ongoing missions will probably be conditional for the existence of a human. Computing errors can produce unexpected results. It would be extremely difficult to guess what a self-teaching AI program can do or decide on the ground if left to your own equipment. There are examples of algorithm qualities, who are trained in intensive learning, come to exceptional, surprising conclusion, which could not even anticipate their creators themselves.

As a moderate period risk, we can withstand a dangerous weapons race among military powers, possibly leading to unaffected arms unaffected arms. In the worst position, non-state actors, terrorists, and/or extremist groups may have access to this technique to be persecuted to minorities (ethnic, religious, or any other) and can be used to pursue their malevolent intentions. The most compelling long -term risk, in the more extreme position, will cause loss of human control over the use of force. Even if robots can ever be more “accurate” and therefore save the life of noncombat, handing, transferring or outsourcing the machines will be a serious sacrifice of war, which will provide it “fundamentally inhuman” (Sharra, 2018, 2018, 285). It often causes moral pain of murder that can question the magnitude of war, increase its political cost, and can act as a barrier to excesses and suffering that wants to stop IHL. By reducing it, it opens the door to accept the restriction of evil because the greater “efficiency” is priced at paying.

Nevertheless, there is indeed still an elusive question. When one talks about “autonomous weapons”, whatever happens in our mind can barely be equal with a single, fully recognizable weapon or device, but basically a process (in which AI plays a central role) that can take any concrete form, can perform cognitive functions, and can carry out complex functions without human intervention. Even for those who are in favor of a pre -procured ban, which can be made a reality, for example, through the closing of a protocol VI of CCW, many questions will be unanswered. Only completely autonomous weapons (human out-loop), just existing weapons, or those who have not been developed yet? Should states exclude defensively used restrictions and ban aggressive weapons only, assuming that in any way it is possible to remove malignant obstacles that make it very difficult to know whether any such weapon is defensive or not? Banning of anti-workers’ weapons (i.e. when people are targets) and except anti-opposing weapons? If by any occasion, during the talks, the concerned state takes a decision in favor of the ban, then who or which body will verify its compliance? AI will undoubtedly make a major verification challenge in all areas of knowledge and civil application. How to monitor their use and how to prevent the misuse of dual use technologies, which use programs and equipment found in systems, equipment and equipment as cell phones, automobiles and virtual reality headsets?

The successful conversation of CCW’s protocol IV left some lessons, often remembered as an example for an earlier ban. Blinding laser weapons were not used earlier in the battlefield, and yet the states decided to prosecute them. Adopting the protocol IV disrupts the development of this type of weapon, already near this technology, thus showing that weapons can be banned if there is a public rebellion against their use (Martine Claus). According to his advocates, preventive sanctions can actually work: Blinding laser weapons has never been used in conflict conditions. In addition, such prohibitions are designed to prevent military use, but do not restrict citizens, peaceful or non-fatal technology applications (laser is often used in drugs and many other civil areas). Another lesson to note is to accept that ICRC and other nongovernmental organizations played an important role in this process, gathering public opinion and spreading expert knowledge on the subject (Mines Action Canada, 2014).¹¹

Additionally, the comparison with chemical weapons is advantageous. The use of chemical weapons was common in armed conflicts like World War I, Vietnam War, and the Iran-Iraq War. It caused extensive harm to both combatants and civilians. However, it was not always the case. In 1993, due to changes in international politics and a new understanding of the strategic interests of major military powers, 193 states signed the Convention on the Prohibition of Chemical Weapons, which now includes all but ninety-eight percent of world population. Despite some ongoing conflicts and controversies, chemical weapons have been labelled as taboo and are being condemned globally. The OPCW, which is based in The Hague, is responsible for upholding the Convention. Applying the same analysis as to autonomous weapons can help us comprehend how political positions and strategic interests are not static; they may actually change over time No proofing e, resulting in previously undreamed outcomes.

In the same vein, what can be learned from nuclear weapons? The atomic bomb was first used in 1945, with horrific human results that are well known. The use of nuclear power for military purposes removed an arms race among superpowers for strategic domination. At the same time, efforts were made to prevent nuclear proliferation from being furious, which was a pioneer for interaction of the Atomic Treaty (NPT) adopted in 1968. The NPT recognized five nuclear powers and banned horizontal spread among other states in expectation. The nuclear power agency (IAEA) was tasked with implementing security measures and inspecting to ensure that non-exercise hesitation was being completed. But the fact that the vertical proliferation continued anyway, with the development of even more sophisticated and destructive bombs, missiles and nuclear warheads, with decades later, a large part of the states, to negotiate a legally binding means to restrict nuclear weapons, forwarding their total elimination. In 2017, the treaty was adopted on the prohibition of nuclear weapons in the United Nations in New York. The treaty prevents the development, testing, production, stockpiling, stationing, stationing, transfer, use and danger of the use of nuclear weapons, thus making illegal under international law to make it illegal that was already present. Can there be a similar development in case of autonomous weapons? Is this really a desirable landscape from the point of view of promoting peace, security and respect for IHL?

Conclusion

Some possible results may arise from consultation under CCW, which can surely decide only on the continuity of a group of government experts, postponing any more comital decision for any longer time. Alternatively, it can give states a mandate to interact in Bayana, so that meetings in Geneva can be suited for more concrete steps towards regulating autonomous weapons. Diplomatic and experts speculate on the viability of legally binding tool regardless of their format and scope. The possibility of a political announcement, some of them, or at least some of them, or the Code of Conduct establishing the rules and principles applied to autonomous weapons have also been identified. Incidentally, we should not dismiss at the point that, if CCW discussions do not give effective results on time and reach a deadlock, then there may be a call for other forums or solutions to measure diplomatic model (Jacobson, 2017, P.5). How will the major AI powers react to these developments? Will the autonomous weapons be forced to impose pre-e-khali ban if they are not on the board? If no rule is in place and IHL is completely ignored, what if the technique moves at the point where the entirely autonomous weapons are in operation for real? Can military powers then embrace the idea of restriction to avoid future vulnerabilities from the attacks from their rivals, in the event of a situation where mutual prevention rendering makes the strategic benefits benefiting the states that are entitled to autonomous weapons?

The serious risk of vengeance by the enemy usually has an effective effect against the actual use of some weapons. This self-restraint system is all more specific when almost certainness is that ventilation is unavoidable. Finally, if some great powers are successful in developing their own AI military capabilities, but do not want other states to do so, will they go to the NPT-like international legally binding equipment to assure their own technical suzerainty once? This paper cannot answer all these questions, but sheer is realizing that some of these scenarios can be seriously endangered, both international law scholars and concerned citizens should be sufficient to put on high warnings. AI Defense technology will continue to move forward, more rapid than ever, even if it appears to be far from the eyes of the general public, more and more is used to enjoy the benefits of technological progress in citizen domains in their everyday life. It is important to engage in this conversation, fully discuss and plan further if we want to know (and perhaps affect) what direction is this technique taking this technique. Will people ever be really safe in a world where machines use deadly force without any human intervention? The future begins to shape.

Footnotes

1. Carlos Batallas, ‘When AI Meets the Laws of War’ (Naked Law 3 October 2024) https://www.ie.edu/insights/articles/when-ai-meets-the-laws-of-war/ accessed 2nd March 2026.

2. Priya Mondal, Dr. Nagesh Sawant, Dr. Ramratan Dhuma, Briding the Gap: Artificial Intelligence In Addressing The Accountability Gap In International Humanitarian Law (2024), https://healthinformaticsjournal.com/index.php/IJMI/article/view/755/713%20, accessed 27th February 2025.

3. Ibid

4. The Fourth Hague Convention of 1907.

5. The Geneva Conventions on 1949, Artificial 36.

6. The additional protocol ,1949 (Geneva conferences), Articles 36

7. The Human Rights Council in 2013,

8. Certain Conventions Weapon on 1981.

9. Carlos Batallas, ‘When AI Meets the Laws of War’ (Naked Law 3 October 2024) accessed 1st March 2025.

10. Ibid.

11. Michael Shriney, ‘International Humanitarian Law ‘(Naked law 11 Jun 2023), https://blog.ipleaders.in/inter national-humanitarian-law-2/, accessed 2nd march 2025.