by Sze Hong LAM
Abstract: Conceptualising the Law of Armed Conflict/International Humanitarian Law (LOAC/IHL) as a legal ‘superposition’, this paper uses it as a ‘hard case’ to challenge three implicit assumptions of the Treaty-Following AI (TFAI) framework: genuine agreement, objective determinacy, and neutral interpretation. The paper advances ‘strategic ambiguity’ as the meta-indeterminacy that explains why LOAC/IHL’s four established indeterminacies — linguistic, legal, structural, and practice — persist. While TFAI could improve compliance with the determinate core, its application to the contested penumbra risks collapsing the discursive space that legitimises LOAC/IHL. The paper advocates ‘contestability by design’, in which TFAI should flag rather than resolve legal indeterminacy
Keywords: Artificial Intelligence; Treaty-Following AI; Alignment; Indeterminacy; International Humanitarian Law; Strategic Ambiguity
