Digital Rights Alliance Protecting Human Rights from Military Technologies, 30 November 2023 derogation is permitted, even in situations of armed conflict or other public emergencies that threaten the life of the nation’.14 26. Countries are utilising LAWS in an unregulated environment to deprive individuals of their right to life. The most documented use of LAWS has been in the Libyan civil war and the Russia-Ukraine War.15 The use of such technology in active conflict zones is greatly concerning. 5.3 Proportionality 27. To ensure that the use of NTMD is done in accordance with international law, its use must be proportionate.16 The perceived advantages must outweigh the harms.17 28. States are increasingly integrating algorithms into law enforcement and national security responses.18 NTMD utilising AI directly impacts proportionately, as autonomous weapons do not have the moral capacity to understand the value of a human life. 29. United Nations Secretary-General, António Guterres, has declared that machines determining proportionality in life-or-death situations is ’politically unacceptable and morally repugnant’.19 30. Although humans are by no means perfect decision makers during conflict, they possess’ compassion and intuition – qualities that algorithms can never possess.20 This poses a unique problem in complying with international rules of war when utilising AI-driven NTMD. 5.4 Necessity 31. The principle of necessity requires that there is no other means possible to eliminate a threat. Any force used must be the minimum amount that is reasonable in the circumstances. 32. It is a questionable whether LAWS will be able to comply with this requirement, not only in the context of conflict, but in policing or counterterrorism. 5.5 Discrimination 33. NTMD, especially those utilising algorithms, may result in discrimination which can lead to tragedy. 34. Algorithmic bias is a phenomenon where an AI outcome results in discrimination.21 When used in civilian contexts there are risks of unlawful discrimination.22 However, when algorithmic bias occurs in NTMD, this can be a matter of life-or-death. 7