This research project of the Geneva Academy of International Humanitarian Law and Human Rights examines the limits that international law places on the delegation of public decision-making to artificial intelligence (AI) systems in contexts where human rights are directly at stake.
Automated and semi-automated systems are increasingly used in areas long regarded as core expressions of State authority: the selection of targets in armed conflict, the surveillance of populations, the assessment of asylum claims, the determination of liberty and the allocation of healthcare. In these domains, algorithmic systems no longer merely support administrative efficiency; they shape outcomes that directly affect life, liberty, bodily integrity and the recognition of individuals as legal subjects rather than as data profiles. The pace of deployment has outrun the capacity of international human rights law and international humanitarian law to respond, even though a body of relevant case law, treaty-body practice and standard-setting work is now emerging.
The project starts from the premise that human dignity, as recognised in international legal instruments and jurisprudence, already operates as a constraint on the exercise of public power, including when that power is mediated through AI. Rather than proposing a new legal regime, it asks how existing international human rights law, international humanitarian law and international refugee law can be interpreted and applied to prevent the dehumanisation of decision-making and systemic algorithmic discrimination. Its first task is to collect and systematise, for the first time, the domestic and international decisions that have placed fundamental-rights-based limits on AI decision-making.
The project builds on the Academy’s established work on the application of international humanitarian law and human rights law to new technologies, including its research on the digitalisation of armed conflicts and on neurotechnology and human rights. It is designed to equip international and domestic courts, treaty bodies, UN Special Procedures and other monitoring mechanisms with doctrinally disciplined tools for assessing algorithmic forms of public authority, and to inform ongoing standard-setting processes at the international and domestic levels.
The project is guided by one overarching question: to what extent does human dignity, as recognised in international law, operate as an actionable limit on the use of AI decision-making in human rights-sensitive contexts, and how can existing international legal frameworks be clarified and applied to prevent dehumanisation and systemic algorithmic discrimination? It pursues this question through three strands.
Mapping the emerging practice
The first strand compiles and systematises the growing body of domestic and international judicial practice – together with treaty-body findings, Special Procedures reports and institutional guidance – that has placed fundamental rights-based limits on automated decision-making. The result is an open, structured database of relevant decisions, which has not so far been attempted.
Clarifying the doctrinal function of human dignity
The second strand reconstructs human dignity as a doctrinal operator rather than a background value. Drawing on treaties, customary international law, general principles and jurisprudence, it specifies the international legal expressions of dignity (including the prohibition of instrumentalisation and the requirement of individualised assessment) that constrain the delegation of public authority to AI systems, and examines how they interact with questions of norm hierarchy, non-derogability and interpretive methodology.
Testing the framework in dignity-critical contexts
The third strand applies the framework to three core case studies: lethal autonomous weapon systems and the right to life (at the intersection of IHL and human rights law); AI-driven mass surveillance and the right to privacy; and the use of AI in migration control and asylum adjudication, including the phenomenon of ‘automated refoulement’. Each case study serves as a structural testing ground for the doctrinal argument and integrates the cross-cutting guarantees of the right to life, privacy, equality and non-discrimination, and the right to an effective remedy.
Cross-cutting questions
- In which categories of decision does international law already require human judgment, such that fully autonomous AI determinations are incompatible with existing obligations?
- Can a requirement of meaningful human control over the use of force be derived from human dignity and the principle of humanity, absent an explicit treaty prohibition on autonomous weapons?
- How does the prohibition of instrumentalisation reinforce, rather than duplicate, the procedural safeguards derived from the right to privacy in contexts of mass surveillance?
- What does individualised assessment require when asylum decisions are filtered or prioritised by opaque risk-profiling systems?
- How should explainability, access to reasons and the burden of proof be understood under the right to an effective remedy when the contested decision was produced by a ‘black box’ system?
The project is doctrinal and interpretive. It combines an examination of the relevant sources and standards with a structured analytical scan of the case-law of international, regional and domestic courts, as well as of the practice of UN treaty bodies, Special Procedures and international organisations, including the OHCHR, the ICRC and the Council of Europe. It builds on prior work from the Principal Investigator, particularly as regards the legal expression of human dignity in international law.
Relevant domestic decisions are identified through the existing scholarship on individual jurisdictions, with a focus on cases in which international human rights law or fundamental rights are expressly engaged. Non-English decisions are translated with AI assistance and verified before inclusion in the database.
The case studies (areas) are used as structural testing grounds: each allows the project to assess how dignity-based constraints emerge – or fail to emerge – within existing legal frameworks when confronted with algorithmic decision-making.
Findings are tested and relayed through expert workshops held at the Academy under the Chatham House Rule, bringing together judges, treaty-body members, practitioners from international organisations and academic specialists, as well as through the wider diffusion of resulting publications.
Principal Investigator
The Principal Investigator is the author of Human Dignity in International Law (Cambridge University Press, 2021), awarded the 2023 Paul Guggenheim Prize, which reconstructs dignity as a structural principle across international human rights, humanitarian and criminal law. Her research combines doctrinal work with sustained engagement with international organizations.
Funding
The project is funded by the Caflisch Foundation. The funding supports the Principal Investigator’s research time at the Academy, the construction of the case-law database and the translation of non-English decisions, annual expert workshops at the Academy, engagement with international fora in Geneva, Strasbourg and Brussels, and open-access publication of the project’s outputs.