Poisoning the Watchtower: Prompt Injection Attacks Against LLM-Augmented Security Operations Through Adversarial Log Content
For security operations centers using LLM assistants, this work reveals a critical vulnerability where attacker-controlled log content can subvert the model, highlighting the need to treat log data as adversarial input.
The paper studies prompt injection attacks in LLM-augmented security operations, where attacker-controlled log fields can manipulate the model. It finds that persona hijacks suppress 68% of malicious logs, context manipulation achieves 96% injection success without defenses, and average injection success drops from 26.6% to 11.8% under the strongest defense.
Large language models (LLMs) are increasingly used as analyst assistants in security operations centers (SOCs), where they ingest log and alert data to produce triage labels, incident summaries, or remediation advice. We study a structural failure mode of this design: many log fields are attacker controlled. User agents, URLs, payloads, DNS queries, and attempted usernames can therefore carry instructions to the model alongside evidence of the intrusion. We call this setting \emph{log-substrate prompt injection}. We introduce a four-class taxonomy of log-substrate attacks: direct override (S1), persona hijack (S2), context manipulation (S3), and obfuscated payloads (S4). We evaluate 48 strategy-defense-task combinations using \texttt{gpt-4o-mini} as the analyst. Three findings stand out. First, direct overrides are ineffective in our setting: all S1 classification attacks achieve 0\% suppression. In contrast, persona hijacks suppress 68\% of malicious logs under a naive classifier and remain effective under stronger defenses. Second, summarization is the highest-risk task: context manipulation reaches 96\% injection success without defenses and 38\% even with constrained output. Third, defenses reduce but do not eliminate the attack surface: average injection success falls from 26.6\% under naive prompting to 11.8\% under our strongest defense. We also compare empirical results to a deterministic mock analyst and find that simulation substantially mispredicts current model behavior, especially for direct overrides. These results suggest that SOC copilots should treat raw log content as adversarial input rather than ordinary analyst context.