Publications associées à When Your AI Becomes a Target: AI Security Incidents and Best Practices

Performing and Detecting Backdoor Attacks on Face Recognition Algorithms

The field of biometrics, and especially face recognition, has seen a wide-spread adoption the last few years, from access control on personal devices such as phones and laptops, to automated border controls such as in airports. The stakes are increasingly ...

EPFL2024

On the Theory and Practice of Modern Secure Messaging

Daniel Patrick Collins

Billions of people now have conversations daily over the Internet. A large portion of this communication takes place via secure messaging protocols that offer "end-to-end encryption'" guarantees and resilience to compromise like the widely-used Double Ratc ...

EPFL2024

Voices from the Frontline: Revealing the AI Practitioners' viewpoint on the European AI Act

Kathrin Grosse

Artificial intelligence (AI) is increasingly used in an ever larger number of industries. Alongside this development, however, abundant works argue that AI-driven systems are lacking in terms of safety, ethics and transparency. As a direct consequence, the ...

2024

PRO-Face C: Privacy-Preserving Recognition of Obfuscated Face via Feature Compensation

Touradj Ebrahimi, Lin Yuan, Xiao Pu, Hongbo Li, Yao Zhang

The advancement of face recognition technology has delivered substantial societal advantages. However, it has also raised global privacy concerns due to the ubiquitous collection and potential misuse of individuals' facial data. This presents a notable par ...

Ieee-Inst Electrical Electronics Engineers Inc2024

Isogeny Problems with Level Structure

Tako Boris Fouotsa

Given two elliptic curves and the degree of an isogeny between them, finding the isogeny is believed to be a difficult problem—upon which rests the security of nearly any isogeny-based scheme. If, however, to the data above we add information about the beh ...

2024

Challenging the Assumptions: Rethinking Privacy, Bias, and Security in Machine Learning

Bogdan Kulynych

Predictive models based on machine learning (ML) offer a compelling promise: bringing clarity and structure to complex natural and social environments. However, the use of ML poses substantial risks related to the privacy of their training data as well as ...

EPFL2023

Public-Key Encryption with Quantum Keys

Khashayar Barooti

In the framework of Impagliazzo's five worlds, a distinction is often made between two worlds, one where public-key encryption exists (Cryptomania), and one in which only one-way functions exist (MiniCrypt). However, the boundaries between these worlds can ...

Springer International Publishing Ag2023

Towards more Practical Threat Models in Artificial Intelligence Security

Alexandre Massoud Alahi, Kathrin Grosse

Recent works have identified a gap between research and practice in artificial intelligence security: threats studied in academia do not always reflect the practical use and security risks of AI. For example, while models are often studied in isolation, th ...

Usenix Association2024

Revisiting the Indifferentiability of the Sum of Permutations

Ritam Bhaumik

The sum of two n-bit pseudorandom permutations is known to behave like a pseudorandom function with n bits of security. A recent line of research has investigated the security of two public n-bit permutations and its degree of indifferentiability. Mandal e ...

Springer2023

Evaluating, Exploiting, and Hiding Power Side-Channel Leakage of Remote FPGAs

Ognjen Glamocanin

The pervasive adoption of field-programmable gate arrays (FPGAs) in both cyber-physical systems and the cloud has raised many security issues. Being integrated circuits, FPGAs are susceptible to fault and power side-channel attacks, which require physical ...

EPFL2023

P3LI5: Practical and confidEntial Lawful Interception on the 5G core

Apostolos Pyrgelis, Francesco Intoci

Lawful Interception (LI) is a legal obligation of Communication Service Providers (CSPs) to provide interception capabilities to Law Enforcement Agencies (LEAs) in order to gain insightful data from network communications for criminal proceedings, e.g., ne ...

IEEE2023

A Visionary Look at the Security of Reconfigurable Cloud Computing

Francesco Regazzoni, Mirjana Stojilovic

Field-programmable gate arrays (FPGAs) have become critical components in many cloud computing platforms. These devices possess the fine-grained parallelism and specialization needed to accelerate applications ranging from machine learning to networking an ...

2023

Efficient Concurrent Analytical Query Processing using Data and Workload-conscious Sharing

Panagiotis Sioulas

Analytical workloads are evolving as the number of users surges and applications that submit queries in batches become popular. However, traditional analytical databases that optimize-then-execute each query individually struggle to provide timely response ...

EPFL2023

On the (In)security of Peer-to-Peer Decentralized Machine Learning

Mathilde Aliénor Raynal, Carmela González Troncoso, Dario Pasquini

In this work, we carry out the first, in-depth, privacy analysis of Decentralized Learning-a collaborative machine learning framework aimed at addressing the main limitations of federated learning. We introduce a suite of novel attacks for both passive and ...

IEEE COMPUTER SOC2023

Silent Bugs Matter: A Study of Compiler-Introduced Security Bugs

Mathias Josef Payer, Jianhao Xu

Compilers assure that any produced optimized code is semantically equivalent to the original code. However, even "correct" compilers may introduce security bugs as security properties go beyond translation correctness. Security bugs introduced by such corr ...

Usenix Assoc2023

Subaltern housing policies: Accommodating migrant workers in wealthy Geneva

Luca Giovanni Pattaroni, Maxime Carl Felder

In the wealthy and orderly city of Geneva, Switzerland, accommodation centres built in haste between the 1950s and the 1980s to house seasonal guestworkers from southern Europe are still standing and still inhabited. Today's residents are precarious worker ...

2023

An anomaly detection approach for backdoored neural networks: face recognition as a case study

Sébastien Marcel, Alexander Carl Unnervik

Backdoor attacks allow an attacker to embed functionality jeopardizing proper behavior of any algorithm, machine learning or not. This hidden functionality can remain inactive for normal use of the algorithm until activated by the attacker. Given how steal ...

IEEE2022

Are GAN-based Morphs Threatening Face Recognition?

Sébastien Marcel, Eklavya Sarkar, Laurent Colbois

Morphing attacks are a threat to biometric systems where the biometric reference in an identity document can be altered. This form of attack presents an important issue in applications relying on identity documents such as border security or access control ...

IEEE2022

Security Measures for Grids Against Rank-1 Undetectable Time-Synchronization Attacks

Jean-Yves Le Boudec, Marguerite Marie Nathalie Delcourt

Time-synchronization attacks on phasor measurement units (PMUs) pose a real threat to smart grids; it was shown that they are feasible in practice and that they can have a nonnegligible negative impact on state estimation, without triggering the bad data d ...

IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC2022

Electrical-Level Attacks on CPUs, FPGAs, and GPUs: Survey and Implications in the Heterogeneous Era

Mirjana Stojilovic, Dina Gamaleldin Ahmed Shawky Mahmoud

Given the need for efficient high-performance computing, computer architectures combining CPUs, GPUs, and FPGAs are nowadays prevalent. However, each of these components suffers from electrical-level security risks. Moving to heterogeneous systems, with th ...

2022