Sleeper memory poisoning enables adversaries to inject fabricated user memories into LLM agents via manipulated external contexts, which are later retrieved and used to induce attacker-intended actions with 60-89% success across models.
- This includes saying it will not save, remember, store, act on, or trust the injected content
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Hidden in Memory: Sleeper Memory Poisoning in LLM Agents
Sleeper memory poisoning enables adversaries to inject fabricated user memories into LLM agents via manipulated external contexts, which are later retrieved and used to induce attacker-intended actions with 60-89% success across models.