FlowGuard: Flow Matching for Identity-Independent Detection of Data-Free Model Stealing Attacks on Energy System Intrusion Detection Systems Conference: ACM Sustainability Week 2026; June 22–25, 2026; Banff, AB, Canada ACM Sustainability Week 2026 (ACM Sustainability Week Companion ’26), June 22–25, 2026, Banff, AB, Canada DOI: 10.1145/3765611.3815137 ISBN: 979-8-4007-2199-1/2026/06 CCS: Security and privacy Software security engineering CCS: Computing methodologies Artificial intelligence CCS: Computing methodologies Neural networks

Maxime Schwarzer, Laurin Holz, Tobias Huerten, Johannes Loevenich, Thies Moehlenhof, Roberto Rigolin F. Lopes, Veit Hagenmeyer · 2026

Read FlowGuard: Flow Matching for Identity-Independent Detection of Data-Free Model Stealing Attacks on Energy System Intrusion Detection Systems Conference: ACM Sustainability Week 2026; June 22–25, 2026; Banff, AB, Canada ACM Sustainability Week 2026 (ACM Sustainability Week Companion ’26), June 22–25, 2026, Banff, AB, Canada DOI: 10.1145/3765611.3815137 ISBN: 979-8-4007-2199-1/2026/06 CCS: Security and privacy Software security engineering CCS: Computing methodologies Artificial intelligence CCS: Computing methodologies Neural networks by Maxime Schwarzer, Laurin Holz, Tobias Huerten, Johannes Loevenich, Thies Moehlenhof, Roberto Rigolin F. Lopes, Veit Hagenmeyer (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-03430.txt.

Description

Schwarzer and co-authors defend energy-system intrusion detectors against model stealing. Attackers can copy a detector by probing it with synthetic traffic. They then craft traffic that slips past the copy and the original. Guards that track user identity fail when attackers spread queries over many accounts. Guards that poison soft labels fail when the detector gives hard labels only.

FlowGuard takes a different route based on flow matching. It flags incoming queries as out of distribution before the detector sees them. Synthetic attack traffic sits on a lower-dimensional manifold than real traffic. A continuous normalizing flow trained on real data scores such queries low. Tests pit FlowGuard against PRADA and FDINet under two known attacks. Attack setups include one client and a spread of 100 fake clients. PRADA's detection fell to zero when the spread changed. FlowGuard held a stable rate in both settings without identity data.

Catalog record

CREATORMaxime Schwarzer, Laurin Holz, Tobias Huerten, Johannes Loevenich, Thies Moehlenhof, Roberto Rigolin F. Lopes, Veit Hagenmeyer
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE24 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2606.03430

Browse the full library · About the Tower of Babel