SwarmReconGuard: Black-Box Detection of Distributed Collective Reconnaissance by Individually Benign-Looking Agent Populations

arXiv:2610.09138 · cs.CR, cs.AI, cs.MA · Submitted 2026-10-06 · Read on arXiv

cs.CR, cs.AI, cs.MA

Submitted: 2026-10-06

Updated: 2026-10-06

Code: https://github.com/vtavakkoli/swarm-recon-guard

License: http://creativecommons.org/licenses/by/4.0/

The gist: Autonomous and agentic clients can distribute reconnaissance across many identities so that each request remains valid, low-rate, and benign-looking while the population collectively acquires broad

Terminology

Abstract

Autonomous and agentic clients can distribute reconnaissance across many identities so that each request remains valid, low-rate, and benign-looking while the population collectively acquires broad system knowledge. We formalize this threat as Distributed Collective Reconnaissance (DCR) and present SwarmReconGuard, a reproducible black-box benchmark in which the defender observes only service-boundary telemetry. The Docker-isolated study evaluates 11 benign and attack behaviors across 10-10,000 virtual identities, comprising 440 test runs and 3,666,300 requests, with complete telemetry integrity. We compare semantic, Gaussian, conditional, graph, kernel, hybrid, and CUSUM-based detectors. Gaussian likelihood-ratio detection achieves 100 % detection with 0 % observed false positives on known attacks but only 3 % on unseen policies. CUSUM yields 36.1 % overall detection at 1.25 % false positives, while hybrid CUSUM reaches 85.7 % detection with 0 % observed false positives at 10,000 identities. Results expose a major policy-generalization gap and motivate exposure-aware, scale-aware defenses.

Sources

Related papers