Lun
20
Lug
Seminari e Convegni
Three Models of AI Regulation
Seminario online Three Models of AI Regulation, tenuto da Ignacio Cofone.
L’evento è svolto nell'ambito del ciclo di seminari “Responsible Artificial Intelligence” (RAI).
Quando: lunedì 20 luglio, ore 16,30
Dove: Online al link: http://tiny.cc/h0v6101
Relatore: Ignacio Cofone, University of Oxford (GB)
Abstract: AI systems allocate credit, price insurance, screen job applicants, triage patients, and inform bail decisions by generating probabilistic inferences at population scale. The resulting harms are group-based, non-material, and produced by long chains of actors, so they slip past legal categories built for individual and tangible injury. Three regulatory traditions have crystallised in response. The individual rights model, exemplified by the GDPR, gives data subjects rights to access, explanation, and objection. The risk-tiered model, exemplified by the EU AI Act, classifies systems before deployment by projected risk. The public infrastructure model, exemplified by algorithmic registries and mandatory audits, treats AI as a governed utility. Each rests on a different theory of the relationship between AI systems and the people they affect, and each carries a blind spot that survives any amount of enforcement. Two of those blind spots reproduce across all three models. Cumulative harm, where thousands of individually unactionable outputs aggregate into systemic injury, escapes every framework. So does the inference lifecycle, which each model leaves ungoverned because all three regulate the data trail and leave the production of inferences unaffected.
Bio: Ignacio Cofone is Professor of Law and Regulation of AI at the University of Oxford, jointly appointed between the Faculty of Law and the Institute for Ethics in AI. His research examines how AI produces inference-based harms that escape existing legal categories, across privacy, data protection, and algorithmic accountability. He is the author of The Privacy Fallacy (Cambridge University Press, 2023) and has advised governments in Canada, the United States, and Europe on AI and privacy regulation.
Per ulteriori informazioni
· Contattare: Professor Daniele Quercia
L’evento è svolto nell'ambito del ciclo di seminari “Responsible Artificial Intelligence” (RAI).
Quando: lunedì 20 luglio, ore 16,30
Dove: Online al link: http://tiny.cc/h0v6101
Relatore: Ignacio Cofone, University of Oxford (GB)
Abstract: AI systems allocate credit, price insurance, screen job applicants, triage patients, and inform bail decisions by generating probabilistic inferences at population scale. The resulting harms are group-based, non-material, and produced by long chains of actors, so they slip past legal categories built for individual and tangible injury. Three regulatory traditions have crystallised in response. The individual rights model, exemplified by the GDPR, gives data subjects rights to access, explanation, and objection. The risk-tiered model, exemplified by the EU AI Act, classifies systems before deployment by projected risk. The public infrastructure model, exemplified by algorithmic registries and mandatory audits, treats AI as a governed utility. Each rests on a different theory of the relationship between AI systems and the people they affect, and each carries a blind spot that survives any amount of enforcement. Two of those blind spots reproduce across all three models. Cumulative harm, where thousands of individually unactionable outputs aggregate into systemic injury, escapes every framework. So does the inference lifecycle, which each model leaves ungoverned because all three regulate the data trail and leave the production of inferences unaffected.
Bio: Ignacio Cofone is Professor of Law and Regulation of AI at the University of Oxford, jointly appointed between the Faculty of Law and the Institute for Ethics in AI. His research examines how AI produces inference-based harms that escape existing legal categories, across privacy, data protection, and algorithmic accountability. He is the author of The Privacy Fallacy (Cambridge University Press, 2023) and has advised governments in Canada, the United States, and Europe on AI and privacy regulation.
Per ulteriori informazioni
· Contattare: Professor Daniele Quercia