DDoD: dual denial of decision attacks on human-AI teams

Benjamin Tag; Niels Van Berkel; Sunny Verma; Benjamin Zi Hao Zhao; Shlomo Berkovsky; Dali Kaafar; Vassilis Kostakos; Olga Ohrimenko

doi:10.1109/MPRV.2022.3218773

DDoD: dual denial of decision attacks on human-AI teams

Benjamin Tag^*, Niels Van Berkel, Sunny Verma, Benjamin Zi Hao Zhao, Shlomo Berkovsky, Dali Kaafar, Vassilis Kostakos, Olga Ohrimenko

^*Corresponding author for this work

Research output: Contribution to journal › Article › peer-review

5 Citations (Scopus)

41 Downloads (Pure)

Abstract

Artificial intelligence (AI) systems have been increasingly used to make decision-making processes faster, more accurate, and more efficient. However, such systems are also at constant risk of being attacked. While the majority of attacks targeting AI-based applications aim to manipulate classifiers or training data and alter the output of an AI model, recently proposed sponge attacks against AI models aim to impede the classifier's execution by consuming substantial resources. In this work, we propose dual denial of decision (DDoD) attacks against collaborative human-AI teams. We discuss how such attacks aim to deplete both computational and human resources, and significantly impair decision-making capabilities. We describe DDoD on human and computational resources and present potential risk scenarios in a series of exemplary domains.

Original language	English
Pages (from-to)	77-84
Number of pages	8
Journal	IEEE Pervasive Computing
Volume	22
Issue number	1
Early online date	27 Feb 2023
DOIs	https://doi.org/10.1109/MPRV.2022.3218773
Publication status	Published - 27 Mar 2023

Bibliographical note

Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher.

Keywords

Artificial intelligence
Data models
Predictive models
Task analysis
Training
Training data
Uncertainty

Access to Document

10.1109/MPRV.2022.3218773Licence: CC BY-NC-ND

Publisher version (open access)Final published version, 454 KBLicence: CC BY-NC-ND

Cite this

@article{c04ad32c8db947ac84abab5d6eea5acc,

title = "DDoD: dual denial of decision attacks on human-AI teams",

abstract = "Artificial intelligence (AI) systems have been increasingly used to make decision-making processes faster, more accurate, and more efficient. However, such systems are also at constant risk of being attacked. While the majority of attacks targeting AI-based applications aim to manipulate classifiers or training data and alter the output of an AI model, recently proposed sponge attacks against AI models aim to impede the classifier's execution by consuming substantial resources. In this work, we propose dual denial of decision (DDoD) attacks against collaborative human-AI teams. We discuss how such attacks aim to deplete both computational and human resources, and significantly impair decision-making capabilities. We describe DDoD on human and computational resources and present potential risk scenarios in a series of exemplary domains. ",

keywords = "Artificial intelligence, Data models, Predictive models, Task analysis, Training, Training data, Uncertainty",

author = "Benjamin Tag and {Van Berkel}, Niels and Sunny Verma and Zhao, {Benjamin Zi Hao} and Shlomo Berkovsky and Dali Kaafar and Vassilis Kostakos and Olga Ohrimenko",

note = "Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher.",

year = "2023",

month = mar,

day = "27",

doi = "10.1109/MPRV.2022.3218773",

language = "English",

volume = "22",

pages = "77--84",

journal = "IEEE Pervasive Computing",

issn = "1536-1268",

publisher = "Institute of Electrical and Electronics Engineers (IEEE)",

number = "1",

}

TY - JOUR

T1 - DDoD

T2 - dual denial of decision attacks on human-AI teams

AU - Tag, Benjamin

AU - Van Berkel, Niels

AU - Verma, Sunny

AU - Zhao, Benjamin Zi Hao

AU - Berkovsky, Shlomo

AU - Kaafar, Dali

AU - Kostakos, Vassilis

AU - Ohrimenko, Olga

N1 - Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher.

PY - 2023/3/27

Y1 - 2023/3/27

N2 - Artificial intelligence (AI) systems have been increasingly used to make decision-making processes faster, more accurate, and more efficient. However, such systems are also at constant risk of being attacked. While the majority of attacks targeting AI-based applications aim to manipulate classifiers or training data and alter the output of an AI model, recently proposed sponge attacks against AI models aim to impede the classifier's execution by consuming substantial resources. In this work, we propose dual denial of decision (DDoD) attacks against collaborative human-AI teams. We discuss how such attacks aim to deplete both computational and human resources, and significantly impair decision-making capabilities. We describe DDoD on human and computational resources and present potential risk scenarios in a series of exemplary domains.

AB - Artificial intelligence (AI) systems have been increasingly used to make decision-making processes faster, more accurate, and more efficient. However, such systems are also at constant risk of being attacked. While the majority of attacks targeting AI-based applications aim to manipulate classifiers or training data and alter the output of an AI model, recently proposed sponge attacks against AI models aim to impede the classifier's execution by consuming substantial resources. In this work, we propose dual denial of decision (DDoD) attacks against collaborative human-AI teams. We discuss how such attacks aim to deplete both computational and human resources, and significantly impair decision-making capabilities. We describe DDoD on human and computational resources and present potential risk scenarios in a series of exemplary domains.

KW - Artificial intelligence

KW - Data models

KW - Predictive models

KW - Task analysis

KW - Training

KW - Training data

KW - Uncertainty

UR - http://www.scopus.com/inward/record.url?scp=85149375614&partnerID=8YFLogxK

U2 - 10.1109/MPRV.2022.3218773

DO - 10.1109/MPRV.2022.3218773

M3 - Article

AN - SCOPUS:85149375614

SN - 1536-1268

VL - 22

SP - 77

EP - 84

JO - IEEE Pervasive Computing

JF - IEEE Pervasive Computing

IS - 1

ER -

DDoD: dual denial of decision attacks on human-AI teams

Abstract

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this