Sleeper Channels and Provenance Gates: Persistent Prompt Injection in Always-on Autonomous AI Agents

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Maloyan, Narek, Namiot, Dmitry
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916009439395840
author Maloyan, Narek
Namiot, Dmitry
author_facet Maloyan, Narek
Namiot, Dmitry
contents Always-on AI agents (OpenClaw, Hermes Agent) run as a single persistent process under the owner's identity, folding messaging, memory, self-authored skills, scheduling, and shell into one authority boundary. This configuration opens what we call \emph{sleeper channels}: an untrusted input to one surface persists as a memory, skill, scheduled job, or filesystem patch, then fires later through a different surface with no attacker present. Two independent axes define the class: persistence substrate and firing-separation. We walk a confused-deputy cron attack end-to-end through OpenClaw at a pinned commit. The defense is tiered (D1, D2, D3), and D2 carries a soundness theorem against seven named deployment invariants. D2 keys on a canonical action-instance digest with one-shot owner attestations, defeating paraphrase laundering, multi-input grant reuse, and replay. A companion artifact ships the gate, a static audit over the vendored source, and a runtime adapter realising five of the ten mediation hooks (H1, H2, H3, H6, H9) around the cron path (42 tests, Node~$\geq{}20$, at \href{https://github.com/maloyan/sleeper-channels}{github.com/maloyan/sleeper-channels}). Empirical evaluation is preregistered as follow-on.
format Preprint
id arxiv_https___arxiv_org_abs_2605_13471
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Sleeper Channels and Provenance Gates: Persistent Prompt Injection in Always-on Autonomous AI Agents
Maloyan, Narek
Namiot, Dmitry
Cryptography and Security
Always-on AI agents (OpenClaw, Hermes Agent) run as a single persistent process under the owner's identity, folding messaging, memory, self-authored skills, scheduling, and shell into one authority boundary. This configuration opens what we call \emph{sleeper channels}: an untrusted input to one surface persists as a memory, skill, scheduled job, or filesystem patch, then fires later through a different surface with no attacker present. Two independent axes define the class: persistence substrate and firing-separation. We walk a confused-deputy cron attack end-to-end through OpenClaw at a pinned commit. The defense is tiered (D1, D2, D3), and D2 carries a soundness theorem against seven named deployment invariants. D2 keys on a canonical action-instance digest with one-shot owner attestations, defeating paraphrase laundering, multi-input grant reuse, and replay. A companion artifact ships the gate, a static audit over the vendored source, and a runtime adapter realising five of the ten mediation hooks (H1, H2, H3, H6, H9) around the cron path (42 tests, Node~$\geq{}20$, at \href{https://github.com/maloyan/sleeper-channels}{github.com/maloyan/sleeper-channels}). Empirical evaluation is preregistered as follow-on.
title Sleeper Channels and Provenance Gates: Persistent Prompt Injection in Always-on Autonomous AI Agents
topic Cryptography and Security
url https://arxiv.org/abs/2605.13471