Herd Accountability of Privacy-Preserving Algorithms: A Stackelberg Game Approach

AI-driven algorithmic systems are increasingly adopted across various sectors, yet the lack of transparency can raise accountability concerns about claimed privacy protection measures. While machine-based audits offer one avenue for addressing these issues, they are often costly and time-consuming....

Full description

Saved in:

Bibliographic Details
Published in:	IEEE transactions on information forensics and security Vol. 20; pp. 2237 - 2251
Main Authors:	Yang, Ya-Ting, Zhang, Tao, Zhu, Quanyan
Format:	Journal Article
Language:	English
Published:	IEEE 2025
Subjects:	accountability Accuracy Algorithm audit Costs Differential privacy Games Government Machine learning algorithms Privacy Protection rational inattention Stackelberg game Training Uniform resource locators
ISSN:	1556-6013, 1556-6021
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Abstract	AI-driven algorithmic systems are increasingly adopted across various sectors, yet the lack of transparency can raise accountability concerns about claimed privacy protection measures. While machine-based audits offer one avenue for addressing these issues, they are often costly and time-consuming. Herd audit, on the other hand, offers a promising alternative by leveraging collective intelligence from end-users. However, the presence of epistemic disparity among auditors, resulting in varying levels of domain expertise and access to relevant knowledge, captured by the rational inattention model, may impact audit assurance. An effective herd audit must establish a credible accountability threat for algorithm developers, incentivizing them not to breach user trust. In this work, our objective is to develop a systematic framework that explores the impact of herd audits on algorithm developers through the lens of the Stackelberg game. Our analysis reveals the importance of easy access to information and the appropriate design of rewards, as they increase the auditors' assurance in the audit process. In this context, herd audit serves as a deterrent to negligent behavior. Therefore, by enhancing herd accountability, herd audit contributes to responsible algorithm development, fostering trust between users and algorithms.
AbstractList	AI-driven algorithmic systems are increasingly adopted across various sectors, yet the lack of transparency can raise accountability concerns about claimed privacy protection measures. While machine-based audits offer one avenue for addressing these issues, they are often costly and time-consuming. Herd audit, on the other hand, offers a promising alternative by leveraging collective intelligence from end-users. However, the presence of epistemic disparity among auditors, resulting in varying levels of domain expertise and access to relevant knowledge, captured by the rational inattention model, may impact audit assurance. An effective herd audit must establish a credible accountability threat for algorithm developers, incentivizing them not to breach user trust. In this work, our objective is to develop a systematic framework that explores the impact of herd audits on algorithm developers through the lens of the Stackelberg game. Our analysis reveals the importance of easy access to information and the appropriate design of rewards, as they increase the auditors' assurance in the audit process. In this context, herd audit serves as a deterrent to negligent behavior. Therefore, by enhancing herd accountability, herd audit contributes to responsible algorithm development, fostering trust between users and algorithms.
Author	Zhu, Quanyan Yang, Ya-Ting Zhang, Tao
Author_xml	– sequence: 1 givenname: Ya-Ting orcidid: 0000-0002-9158-1722 surname: Yang fullname: Yang, Ya-Ting email: yy4348@nyu.edu organization: Department of Electrical and Computer Engineering, New York University, Brooklyn, NY, USA – sequence: 2 givenname: Tao orcidid: 0000-0002-1454-4645 surname: Zhang fullname: Zhang, Tao email: tz636@nyu.edu organization: Department of Electrical and Computer Engineering, New York University, Brooklyn, NY, USA – sequence: 3 givenname: Quanyan orcidid: 0000-0002-0008-2953 surname: Zhu fullname: Zhu, Quanyan email: qz494@nyu.edu organization: Department of Electrical and Computer Engineering, New York University, Brooklyn, NY, USA
BookMark	eNpNkNtqwkAYhJdioWr7AIVe7AvE7jHZ9C5IPYBQi_Y6_Nn80W1jIrup4NurKKVXMxczw_ANSK9pGyTkmbMR5yx9Xc8nq5FgQo-kVkzq5I70udZxFDPBe3-eywcyCOGbMaV4bPrkc4a-pJm17W_TQeFq1x1pW9Gldwewx2jpMaA_uGZDs3rTetdtd-GNZnTVgf3BukC_oVPYIc32e9-C3T6S-wrqgE83HZKvyft6PIsWH9P5OFtElivTRSiQQVyyUkiRmvMdK-JCohYAIIoqASMVmjhVlU1QKW0Eh9JaEGeF1Fo5JPy6a30bgscq33u3A3_MOcsvTPILk_zCJL8xOXderh2HiP_yJklZYuQJIWBgMg
CODEN	ITIFA6
Cites_doi	10.1007/s12599-010-0114-8 10.1016/S0304-3932(03)00029-1 10.1145/3449148 10.1016/j.ejor.2019.05.012 10.1145/3139256 10.1037/0033-295X.111.4.1036 10.1007/s10796-012-9350-4 10.1109/MWC.2016.7721739 10.1145/3243734.3243863 10.4324/9781315212043-31 10.1109/CIC.2018.00054 10.1109/JIOT.2020.2996229 10.1109/TIFS.2021.3118886 10.1145/2480741.2480742 10.1017/S0265052522000085 10.1002/9781119723950.ch2 10.1109/MIC.2012.68 10.1007/978-3-031-30709-6 10.1109/LCSYS.2021.3132801 10.1016/j.procs.2019.08.098 10.1002/0471200611 10.1016/j.cogsys.2007.07.001 10.1287/opre.2013.1235 10.4135/9781452230597.n2 10.1109/JSAC.2012.121214 10.1109/CDC51059.2022.9993423 10.1145/3243734.3243818 10.1145/3057268 10.3390/s22031032 10.1111/j.1099-1123.2009.00413.x 10.1093/cybsec/tyae009 10.1007/978-3-319-47413-7_19 10.1257/aer.20140117 10.1016/j.cose.2018.01.005 10.1109/TIFS.2019.2955891 10.1007/978-3-540-79228-4_1 10.1007/978-3-319-68711-7_7 10.3390/g15040028 10.1038/s41598-016-0011-6 10.1016/j.red.2012.03.003 10.1257/aer.20130047 10.1007/978-3-030-66065-9_9 10.3390/su12239827 10.4108/icst.collaboratecom.2012.250499 10.1007/978-3-031-50670-3_18 10.1016/j.jcorpfin.2020.101588 10.1111/j.1540-6261.2007.01263.x 10.1093/acprof:oso/9780198237907.001.0001 10.1257/jel.20211524 10.1007/s11831-024-10095-6 10.1145/3159652.3159654
ContentType	Journal Article
DBID	97E RIA RIE AAYXX CITATION
DOI	10.1109/TIFS.2025.3540357
DatabaseName	IEEE All-Society Periodicals Package (ASPP) 2005–Present IEEE All-Society Periodicals Package (ASPP) 1998–Present IEEE/IET Electronic Library (IEL) (UW System Shared) CrossRef
DatabaseTitle	CrossRef
DatabaseTitleList
Database_xml	– sequence: 1 dbid: RIE name: IEEE/IET Electronic Library (IEL) (UW System Shared) url: https://ieeexplore.ieee.org/ sourceTypes: Publisher
DeliveryMethod	fulltext_linktorsrc
Discipline	Engineering Government Computer Science
EISSN	1556-6021
EndPage	2251
ExternalDocumentID	10_1109_TIFS_2025_3540357 10879078
Genre	orig-research
GroupedDBID	0R~ 29I 4.4 5GY 5VS 6IK 97E AAJGR AARMG AASAJ AAWTH ABAZT ABQJQ ABVLG ACGFS ACIWK AENEX AETIX AGQYO AGSQL AHBIQ AKJIK AKQYR ALMA_UNASSIGNED_HOLDINGS ATWAV BEFXN BFFAM BGNUA BKEBE BPEOZ CS3 DU5 EBS EJD HZ~ IFIPE IPLJI JAVBF LAI M43 O9- OCL P2P PQQKQ RIA RIE RNS AAYXX CITATION
ID	FETCH-LOGICAL-c148t-e2e0a6d0d23298044c26b3e52aaa2bf7a834e8694fc7e445821adcca221aa9cc3
IEDL.DBID	RIE
ISICitedReferencesCount	0
ISICitedReferencesURI	http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001435443500007&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN	1556-6013
IngestDate	Sat Nov 29 08:14:10 EST 2025 Wed Aug 27 01:49:28 EDT 2025
IsPeerReviewed	true
IsScholarly	true
Language	English
License	https://ieeexplore.ieee.org/Xplorehelp/downloads/license-information/IEEE.html https://doi.org/10.15223/policy-029 https://doi.org/10.15223/policy-037
LinkModel	DirectLink
MergedId	FETCHMERGED-LOGICAL-c148t-e2e0a6d0d23298044c26b3e52aaa2bf7a834e8694fc7e445821adcca221aa9cc3
ORCID	0000-0002-9158-1722 0000-0002-1454-4645 0000-0002-0008-2953
PageCount	15
ParticipantIDs	ieee_primary_10879078 crossref_primary_10_1109_TIFS_2025_3540357
PublicationCentury	2000
PublicationDate	20250000 2025-00-00
PublicationDateYYYYMMDD	2025-01-01
PublicationDate_xml	– year: 2025 text: 20250000
PublicationDecade	2020
PublicationTitle	IEEE transactions on information forensics and security
PublicationTitleAbbrev	TIFS
PublicationYear	2025
Publisher	IEEE
Publisher_xml	– name: IEEE
References	ref13 ref57 ref12 ref56 ref15 ref59 ref14 ref58 ref53 ref52 ref11 ref55 Guszcza (ref1) 2018 ref10 ref54 ref16 ref51 ref50 Tramer (ref20) 2022 ref46 ref45 ref48 ref47 ref42 ref41 ref44 Nasr (ref23) ref49 ref8 ref9 ref4 ref3 ref6 ref5 ref40 ref35 ref34 ref37 ref36 ref31 ref30 ref33 ref32 Steinke (ref22); 36 ref2 ref39 ref38 Dynan (ref43) Jagielski (ref18); 33 ref24 ref26 (ref62) 2001 Mittelstadt (ref7) 2016; 10 ref25 ref28 ref27 Lu (ref19); 35 ref29 Stevens (ref21) 2022 ref60 ref61 Ge (ref17) 2022
References_xml	– ident: ref9 doi: 10.1007/s12599-010-0114-8 – ident: ref41 doi: 10.1016/S0304-3932(03)00029-1 – ident: ref6 doi: 10.1145/3449148 – ident: ref52 doi: 10.1016/j.ejor.2019.05.012 – ident: ref24 doi: 10.1145/3139256 – ident: ref40 doi: 10.1037/0033-295X.111.4.1036 – ident: ref25 doi: 10.1007/s10796-012-9350-4 – ident: ref34 doi: 10.1109/MWC.2016.7721739 – start-page: 16 volume-title: Proc. Financial Innov. Real Economy’ Conf. Sponsored Center Study Innov. Productiv. ident: ref43 article-title: Financial innovation and the great moderation: What do household data say? – start-page: 1631 volume-title: Proc. 32nd USENIX Secur. Symp. ident: ref23 article-title: Tight auditing of differentially private machine learning – ident: ref4 doi: 10.1145/3243734.3243863 – start-page: 19 volume-title: Classical Detection and Estimation Theory year: 2001 ident: ref62 – ident: ref11 doi: 10.4324/9781315212043-31 – ident: ref60 doi: 10.1109/CIC.2018.00054 – ident: ref36 doi: 10.1109/JIOT.2020.2996229 – ident: ref58 doi: 10.1109/TIFS.2021.3118886 – ident: ref12 doi: 10.1145/2480741.2480742 – volume: 33 start-page: 22205 volume-title: Proc. Adv. Neural Inf. Process. Syst. ident: ref18 article-title: Auditing differentially private machine learning: How private is private SGD? – ident: ref27 doi: 10.1017/S0265052522000085 – ident: ref13 doi: 10.1002/9781119723950.ch2 – ident: ref32 doi: 10.1109/MIC.2012.68 – year: 2022 ident: ref17 article-title: Accountability and insurance in IoT supply chain publication-title: arXiv:2201.11855 – ident: ref39 doi: 10.1007/978-3-031-30709-6 – year: 2022 ident: ref21 article-title: Backpropagation clipping for deep learning with differential privacy publication-title: arXiv:2202.05089 – ident: ref5 doi: 10.1109/LCSYS.2021.3132801 – ident: ref47 doi: 10.1016/j.procs.2019.08.098 – ident: ref61 doi: 10.1002/0471200611 – volume: 35 start-page: 4165 volume-title: Proc. Adv. Neural Inf. Process. Syst. ident: ref19 article-title: A general framework for auditing differentially private machine learning – ident: ref38 doi: 10.1016/j.cogsys.2007.07.001 – volume: 10 start-page: 12 year: 2016 ident: ref7 article-title: Automation, algorithms, and politics-auditing for transparency in content personalization systems publication-title: Int. J. Commun. – ident: ref33 doi: 10.1287/opre.2013.1235 – ident: ref37 doi: 10.4135/9781452230597.n2 – ident: ref57 doi: 10.1109/JSAC.2012.121214 – ident: ref59 doi: 10.1109/CDC51059.2022.9993423 – ident: ref3 doi: 10.1145/3243734.3243818 – ident: ref46 doi: 10.1145/3057268 – ident: ref50 doi: 10.3390/s22031032 – ident: ref8 doi: 10.1111/j.1099-1123.2009.00413.x – ident: ref54 doi: 10.1093/cybsec/tyae009 – ident: ref55 doi: 10.1007/978-3-319-47413-7_19 – ident: ref16 doi: 10.1257/aer.20140117 – ident: ref53 doi: 10.1016/j.cose.2018.01.005 – ident: ref56 doi: 10.1109/TIFS.2019.2955891 – ident: ref2 doi: 10.1007/978-3-540-79228-4_1 – ident: ref51 doi: 10.1007/978-3-319-68711-7_7 – ident: ref49 doi: 10.3390/g15040028 – ident: ref29 doi: 10.1038/s41598-016-0011-6 – ident: ref42 doi: 10.1016/j.red.2012.03.003 – ident: ref15 doi: 10.1257/aer.20130047 – ident: ref26 doi: 10.1007/978-3-030-66065-9_9 – volume-title: Why We Need to Audit Algorithms year: 2018 ident: ref1 – ident: ref30 doi: 10.3390/su12239827 – ident: ref35 doi: 10.4108/icst.collaboratecom.2012.250499 – volume: 36 start-page: 49268 volume-title: Proc. Adv. Neural Inf. Process. Syst. ident: ref22 article-title: Privacy auditing with one (1) training run – ident: ref14 doi: 10.1007/978-3-031-50670-3_18 – ident: ref28 doi: 10.1016/j.jcorpfin.2020.101588 – ident: ref44 doi: 10.1111/j.1540-6261.2007.01263.x – ident: ref10 doi: 10.1093/acprof:oso/9780198237907.001.0001 – ident: ref45 doi: 10.1257/jel.20211524 – ident: ref48 doi: 10.1007/s11831-024-10095-6 – year: 2022 ident: ref20 article-title: Debugging differential privacy: A case study for privacy auditing publication-title: arXiv:2202.12219 – ident: ref31 doi: 10.1145/3159652.3159654
SSID	ssj0044168
Score	2.4160025
Snippet	AI-driven algorithmic systems are increasingly adopted across various sectors, yet the lack of transparency can raise accountability concerns about claimed...
SourceID	crossref ieee
SourceType	Index Database Publisher
StartPage	2237
SubjectTerms	accountability Accuracy Algorithm audit Costs Differential privacy Games Government Machine learning algorithms Privacy Protection rational inattention Stackelberg game Training Uniform resource locators
Title	Herd Accountability of Privacy-Preserving Algorithms: A Stackelberg Game Approach
URI	https://ieeexplore.ieee.org/document/10879078
Volume	20
WOSCitedRecordID	wos001435443500007&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText	1
inHoldings	1
isFullTextHit
isPrint
journalDatabaseRights	– providerCode: PRVIEE databaseName: IEEE/IET Electronic Library (IEL) (UW System Shared) customDbUrl: eissn: 1556-6021 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0044168 issn: 1556-6013 databaseCode: RIE dateStart: 20060101 isFulltext: true titleUrlDefault: https://ieeexplore.ieee.org/ providerName: IEEE
link	http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3NT8IwFH8R4gEPoogRv9KDJ5PiWLd19bYYES8EIybcltK-IYkwMwYJ_73tNiIXD562LFuyvNf31df3-wHcqWliooYKKNMOpx6XPhXSn1IftWeWjxRMqYJsgg-H4WQiRtWwejELg4jF4TPs2tuil69TtbZbZcbCQ26KubAGNc6Dclhr53ZNWC_n3nw_oKbKYFULs-eIh_Fr_92Ugq7ftbsczIaivSC0x6pSBJV-85-_cwLHVfZIolLdp3CAyxY0d8wMpDLUFhztwQy2oPFLqHsGbwPMNKlIIkqQ7i1JEzLK5hupttSeybD-Yzkj0dcszeb552L1SCJi0lJj8QUkFnmRCyRRBUfeho_-8_hpQCteBapM8ZNTdNGRgXa0yaZEaCSn3GDK0HellO404TJkHoaB8BLF0bONtZ7URtOuuUqhFDuH-jJd4gUQ1wmURuQidJmFWpMME6GM_9SBRM5lB-53go6_S_iMuCg7HBFbrcRWK3GllQ60rZD3Xizle_nH8yto2M_LDZFrqOfZGm_gUG3y-Sq7LVbHD8ssuFc
linkProvider	IEEE
linkToHtml	http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3NS8MwFH_oFJwHp3Pi_MzBkxCtSds03oY4N5xDcYK3kiWvOnCbdB-w_96k7XAXD55aSinlvbyvvLzfD-BC9xMbNXRIufEE9YUKqFRBnwZofLt8lORaZ2QTotuN3t_lczGsns3CIGJ2-Ayv3G3WyzdjPXNbZdbCI2GLuWgdNgLfZ14-rrV0vDaw55NvQRBSW2fwool548nrXrv5aotBFly5fQ7ugtFKGFrhVcnCSrPyzx_ahZ0ifySNXOF7sIajKlSW3AykMNUqbK8ADVah_Eupuw8vLUwNKWgicpjuBRkn5DkdzJVeUHcqw3mQ0QdpfH2M08H0czi5JQ1iE1Nr8xkoFnlQQySNApC8Bm_N-95dixbMClTb8mdKkaGnQuMZm0_JyEpOs7DPMWBKKdZPhIq4j1Eo_UQL9F1r7UYZq2tmr0pqzQ-gNBqP8BAI80JtEIWMGHdga4pjIrX1oCZUKISqw-VS0PF3DqARZ4WHJ2OnldhpJS60UoeaE_LKi7l8j_54fg5brd5TJ-60u4_HUHafyrdHTqA0TWd4Cpt6Ph1M0rNspfwAj9a7ng
openUrl	ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Herd+Accountability+of+Privacy-Preserving+Algorithms%3A+A+Stackelberg+Game+Approach&rft.jtitle=IEEE+transactions+on+information+forensics+and+security&rft.au=Yang%2C+Ya-Ting&rft.au=Zhang%2C+Tao&rft.au=Zhu%2C+Quanyan&rft.date=2025&rft.issn=1556-6013&rft.eissn=1556-6021&rft.volume=20&rft.spage=2237&rft.epage=2251&rft_id=info:doi/10.1109%2FTIFS.2025.3540357&rft.externalDBID=n%2Fa&rft.externalDocID=10_1109_TIFS_2025_3540357
thumbnail_l	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1556-6013&client=summon
thumbnail_m	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1556-6013&client=summon
thumbnail_s	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1556-6013&client=summon