Policy Brief No. 197 — February 2025
Responsible AI and Civilian Protection in Armed Conflict Daniel R. Mahanty and Kailee Hilt
Key Points → While the Political Declaration on Responsible Military Use of Artificial Intelligence and Autonomy holds promise, its supporters should place greater emphasis on how the implementation of its principles will lead to better protection of civilians in armed conflict, especially when combined with other measures not limited to the use of artificial intelligence (AI) and autonomy. → This policy brief argues that the responsible use invoked by the declaration should not result in only marginally better protection of civilians (PoC)1 outcomes than “irresponsible” use, but should instead achieve markedly better ones. → Giving meaning to the declaration’s implied PoC content depends on whether the expansion of its membership and stewardship of the process raises the ceiling or lowers the floor for responsible use. → National and multilateral efforts to promote the responsible military use of AI should be connected to a renewed commitment among all states to mitigate harm to civilians resulting from all military operations, not only those that involve the use of AI.
1
This brief uses the term “protection of civilians” to mean the measures taken by states to prevent, minimize and address harm resulting from their own military operations (including operations involving allies and partners). This use most closely approximates the concepts of combatant PoC or civilian harm mitigation, rather than its use in the context of peacekeeping or atrocities prevention.
Introduction In the last annual report (2023) on the Protection of Civilians in Armed Conflict, the UN Secretary-General described the situation for civilians in armed conflict in the previous year as “resoundingly grim.” Over the course of 2022, thousands of civilians died and millions more suffered from the impact of wars.2 By the time of the report’s release in the spring of 2023, it appeared that the scope and severity of civilian harm in 2024 would be even worse. The report also arrived amid a growing crisis of public confidence in international humanitarian law, caused by the pervasive and wanton disregard for its principles exhibited by some states and the tepid compliance modelled by others — with a dearth of meaningful accountability on both sides. All the while, disparate levels of concern among Western states for civilians in Gaza, Sudan and Ukraine had led to charges of double standards and hypocrisy. Against this backdrop (a mere six days after the release of the report), the United States published an updated version of its Political Declaration on Responsible Military Use of Artificial Intelligence and Autonomy, replete with an expanded list of supporting states (US Department of State 2023). While the political declaration makes only one explicit reference to the goal of protecting civilians (and that made only in the context of the use of AI and autonomy), several of its elements serve that goal. By committing to use AI in adherence with international law, the declaration’s signatories agree to abide by those
2
See https://acleddata.com/data/#/dashboard.