P216. Governing Algorithmic State Power: Legal and Policy Frameworks for AI in Criminal Justice and Public Administration
Regulation and Regulatory Governance / Multilevel Governance and Local PolicyCATEGORISATION
KEYWORDS
GENERAL OBJECTIVES, RESEARCH QUESTIONS AND SCIENTIFIC RELEVANCE
Governments are adopting artificial intelligence (AI) and automated decision-making (ADM) in policing, criminal justice, welfare and regulatory enforcement. Predictive policing, facial recognition, risk scores in bail and sentencing, and algorithmic fraud detection now shape decisions about liberty, livelihoods and access to justice, often without transparency or redress. The Dutch childcare benefits scandal, Australia’s Robodebt scheme, the SyRI judgment and State v. Loomis show how automated state action can cause systemic harm, and how unevenly courts and governments respond. In India and across the Global South, facial recognition in policing and algorithmic welfare targeting are outpacing legal safeguards.
Public administration research traces a shift from street-level to system-level bureaucracy, in which discretion moves into software design (Bovens & Zouridis 2002). Legal scholars call for technological due process (Citron 2008). Studies of the digital welfare state find harms fall hardest on poor and marginalized people (Eubanks 2018; Alston 2019), and work on accountable AI shows how opacity and proprietary secrecy weaken accountability (Busuioc 2021). This literature centres on North American and European cases, and policy debate often narrows to “AI ethics” or fairness. Legal frameworks such as constitutional rights, data protection law, the EU AI Act and the Council of Europe’s Framework Convention on AI are rarely studied alongside policy frameworks like national AI strategies, impact assessments, procurement standards and oversight bodies. Their fit with jurisdictions that have different administrative traditions and histories of discriminatory policing, including caste-based and colonial-era criminalization still remains under-examined.
The panel asks how AI and ADM change the exercise of state power in criminal justice and public administration, and which combinations of legal and policy instruments can hold it to account. Some of the research questions and hypotheses are mentioned below:
i) Accountability gaps widen when public bodies buy systems from private vendors under commercial confidentiality, spreading responsibility across “many hands”.
ii) Systems built on administrative and policing data reproduce the selectivity of past state practice, concentrating harm on groups already over-policed or under-served along lines of gender, caste and class.
iii)Safeguards such as mandatory human review tend to become rubber stamps (Green 2022), so procedural rights in law do not ensure meaningful contestation in practice.
iv)Where independent regulators are weak, courts and civil society become the main sites of algorithmic accountability, with uneven results.
The panel treats algorithmic decision-making as public policy whose design, procurement, implementation and oversight are contestable political choices. It aims to connect law, public administration and policy studies across regions, with particular attention to the Global South.
CALL FOR PAPERS
We invite theoretical, doctrinal, empirical and comparative papers on the use of AI and automated decision-making by public authorities, particularly in criminal justice, policing, welfare and regulatory enforcement. Papers may discuss (but not limited to) the following themes:
(i) The adequacy of constitutional, human rights, administrative and data protection law in regulating AI-driven state action, including reason-giving, explainability, privacy and proportionality
(ii) The design and implementation of policy instruments such as national AI strategies, algorithmic impact assessments, procurement rules, algorithm registers and audit requirements, and whether they change outcomes
(iii) The differential impact of algorithmic systems on marginalized groups along lines of gender, caste, class, religion or migration status
(iv) Accountability for algorithmic harm through courts, administrative bodies (ombudsmen, data protection authorities, auditors) or civil society (strategic litigation, investigative journalism, community audits)
(v) How governance models such as the EU AI Act travel to the Global South, and how AI interacts with biometric identity systems and other digital public infrastructure
(vi) AI in criminal law and law enforcement, including predictive policing, facial recognition, risk assessment in bail, sentencing and parole, and AI-assisted forensic evidence, and its implications for due process and the presumption of innocence.
We particularly welcome contributions from Global South jurisdictions, where deployment is expanding quickly and research remains limited, and papers that combine legal and policy analysis.
Papers that follow a specific system through the policy cycle, from procurement to implementation and contestation by affected people, are especially valuable. Single-country case studies, cross-national comparisons and normative papers proposing workable models of regulation, oversight or redress are welcome. We encourage submissions from early-career researchers and scholar-practitioner teams.
Abstracts should state the research question, the jurisdiction or case studied, the method, and expected or preliminary findings. Papers should make clear how their findings speak to legal or policy design, so the panel can compare cases and identify lessons for regulators, courts, oversight bodies and civil society.
EXPORT PDF (FULL)