Distributed Agentic Attacks: A Proposed Threat Model for Large-Scale Distributed Offensive Agency

This record contains only Distributed Agentic Attacks: A Proposed Threat Model for Large-Scale Distributed Offensive Agency, Publication Paper v6.4.3. The paper proposes Distributed Agentic Attacks (DAA) as a provisional campaign-level threat-model subtype in which consequential tactical decision authority is distributed across multiple artificial decision principals operating within one coherent offensive campaign. It distinguishes this authority-placement construct from botnets, ordinary distributed infrastructure, conventional automation, agent teams, swarms, and cross-context monitoring fragmentation. Based on the documented targeted literature review, the paper advances a bounded priority claim: this is the first known research program to define this authority-placement construct explicitly and experimentally separate three propositions—exercised distributed authority, architecture-neutral scaled load, and an incremental Distributed-over-Central defender-burden premium. This is not a claim to the first multi-agent cyberattack, offensive agent team, swarm, or monitoring-fragmentation study. The reported terminal-symbolic experiments found that separately assigned decision principals could exercise bounded consequential choices and that scaled accepted activity could overload a fixed symbolic defender. However, the registered analyses did not support a repeatable additional burden attributable specifically to distributing matched decision authority. Study A’s null result was also limited by measurement ceilings. DAA therefore remains a falsifiable provisional construct rather than an empirically established real-world threat class. This is a preprint and has not been peer reviewed. The companion Research Dossier and Source-Provenance Record v4.4.3 and Experimental Evidence and Reproducibility Record v1.4.2 are deposited separately. The research was independently conducted and self-funded by Mario Oliva, with no declared competing interests.