Research · curated 21 Jul 2026
Your Agent's Memories Are Not Its Own: Forged Reasoning Attacks on LLM Agent Memory and Defenses
First reported · updated · 2 reports arxiv.org
Coverage timeline
Why it matters
FARMA demonstrates that the integrity of an agent's own stored reasoning history is a distinct, exploitable attack surface that existing memory-poisoning defenses fail to cover, forcing defenders to protect reasoning traces and not just retrieved content.
Researchers at Penn State introduce FARMA (Forged Amplifying Rationale Memory Attack), which poisons an LLM agent's remembered reasoning traces rather than its factual knowledge, using evasive language to bypass keyword filters and self-referential reinforcement to defeat consensus-based defenses, achieving up to 100% attack success including against A-MemGuard. They also propose SENTINEL, a layered defense whose Reasoning Guard structurally analyzes memory entries for forgery, reducing attack success to as low as 0% with no false positives across 326 benign traces.