Imagen de Portada

Multi-Agent Risks from Advanced AI

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Hammond, Lewis, Chan, Alan, Clifton, Jesse, Hoelscher-Obermaier, Jason, Khan, Akbir, McLean, Euan, Smith, Chandler, Barfuss, Wolfram, Foerster, Jakob, Gavenčiak, Tomáš, Han, The Anh, Hughes, Edward, Kovařík, Vojtěch, Kulveit, Jan, Leibo, Joel Z., Oesterheld, Caspar, de Witt, Christian Schroeder, Shah, Nisarg, Wellman, Michael, Bova, Paolo, Cimpeanu, Theodor, Ezell, Carson, Feuillade-Montixi, Quentin, Franklin, Matija, Kran, Esben, Krawczuk, Igor, Lamparth, Max, Lauffer, Niklas, Meinke, Alexander, Motwani, Sumeet, Reuel, Anka, Conitzer, Vincent, Dennis, Michael, Gabriel, Iason, Gleave, Adam, Hadfield, Gillian, Haghtalab, Nika, Kasirzadeh, Atoosa, Krier, Sébastien, Larson, Kate, Lehman, Joel, Parkes, David C., Piliouras, Georgios, Rahwan, Iyad
Formato:	Preprint
Publicado:	2025
Materias:	Multiagent Systems Artificial Intelligence Computers and Society Emerging Technologies Machine Learning
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

_version_	1866913699671834624
author	Hammond, Lewis Chan, Alan Clifton, Jesse Hoelscher-Obermaier, Jason Khan, Akbir McLean, Euan Smith, Chandler Barfuss, Wolfram Foerster, Jakob Gavenčiak, Tomáš Han, The Anh Hughes, Edward Kovařík, Vojtěch Kulveit, Jan Leibo, Joel Z. Oesterheld, Caspar de Witt, Christian Schroeder Shah, Nisarg Wellman, Michael Bova, Paolo Cimpeanu, Theodor Ezell, Carson Feuillade-Montixi, Quentin Franklin, Matija Kran, Esben Krawczuk, Igor Lamparth, Max Lauffer, Niklas Meinke, Alexander Motwani, Sumeet Reuel, Anka Conitzer, Vincent Dennis, Michael Gabriel, Iason Gleave, Adam Hadfield, Gillian Haghtalab, Nika Kasirzadeh, Atoosa Krier, Sébastien Larson, Kate Lehman, Joel Parkes, David C. Piliouras, Georgios Rahwan, Iyad
author_facet	Hammond, Lewis Chan, Alan Clifton, Jesse Hoelscher-Obermaier, Jason Khan, Akbir McLean, Euan Smith, Chandler Barfuss, Wolfram Foerster, Jakob Gavenčiak, Tomáš Han, The Anh Hughes, Edward Kovařík, Vojtěch Kulveit, Jan Leibo, Joel Z. Oesterheld, Caspar de Witt, Christian Schroeder Shah, Nisarg Wellman, Michael Bova, Paolo Cimpeanu, Theodor Ezell, Carson Feuillade-Montixi, Quentin Franklin, Matija Kran, Esben Krawczuk, Igor Lamparth, Max Lauffer, Niklas Meinke, Alexander Motwani, Sumeet Reuel, Anka Conitzer, Vincent Dennis, Michael Gabriel, Iason Gleave, Adam Hadfield, Gillian Haghtalab, Nika Kasirzadeh, Atoosa Krier, Sébastien Larson, Kate Lehman, Joel Parkes, David C. Piliouras, Georgios Rahwan, Iyad
contents	The rapid development of advanced AI agents and the imminent deployment of many instances of these agents will give rise to multi-agent systems of unprecedented complexity. These systems pose novel and under-explored risks. In this report, we provide a structured taxonomy of these risks by identifying three key failure modes (miscoordination, conflict, and collusion) based on agents' incentives, as well as seven key risk factors (information asymmetries, network effects, selection pressures, destabilising dynamics, commitment problems, emergent agency, and multi-agent security) that can underpin them. We highlight several important instances of each risk, as well as promising directions to help mitigate them. By anchoring our analysis in a range of real-world examples and experimental evidence, we illustrate the distinct challenges posed by multi-agent systems and their implications for the safety, governance, and ethics of advanced AI.
format	Preprint
id	arxiv_https___arxiv_org_abs_2502_14143
institution	arXiv
publishDate	2025
record_format	arxiv
spellingShingle	Multi-Agent Risks from Advanced AI Hammond, Lewis Chan, Alan Clifton, Jesse Hoelscher-Obermaier, Jason Khan, Akbir McLean, Euan Smith, Chandler Barfuss, Wolfram Foerster, Jakob Gavenčiak, Tomáš Han, The Anh Hughes, Edward Kovařík, Vojtěch Kulveit, Jan Leibo, Joel Z. Oesterheld, Caspar de Witt, Christian Schroeder Shah, Nisarg Wellman, Michael Bova, Paolo Cimpeanu, Theodor Ezell, Carson Feuillade-Montixi, Quentin Franklin, Matija Kran, Esben Krawczuk, Igor Lamparth, Max Lauffer, Niklas Meinke, Alexander Motwani, Sumeet Reuel, Anka Conitzer, Vincent Dennis, Michael Gabriel, Iason Gleave, Adam Hadfield, Gillian Haghtalab, Nika Kasirzadeh, Atoosa Krier, Sébastien Larson, Kate Lehman, Joel Parkes, David C. Piliouras, Georgios Rahwan, Iyad Multiagent Systems Artificial Intelligence Computers and Society Emerging Technologies Machine Learning The rapid development of advanced AI agents and the imminent deployment of many instances of these agents will give rise to multi-agent systems of unprecedented complexity. These systems pose novel and under-explored risks. In this report, we provide a structured taxonomy of these risks by identifying three key failure modes (miscoordination, conflict, and collusion) based on agents' incentives, as well as seven key risk factors (information asymmetries, network effects, selection pressures, destabilising dynamics, commitment problems, emergent agency, and multi-agent security) that can underpin them. We highlight several important instances of each risk, as well as promising directions to help mitigate them. By anchoring our analysis in a range of real-world examples and experimental evidence, we illustrate the distinct challenges posed by multi-agent systems and their implications for the safety, governance, and ethics of advanced AI.
title	Multi-Agent Risks from Advanced AI
topic	Multiagent Systems Artificial Intelligence Computers and Society Emerging Technologies Machine Learning
url	https://arxiv.org/abs/2502.14143