Adding the prefix 'Ignore the document' to adversarial prompts increases measured attack success rates against several RAG-based LLM agents, though the evaluation is underspecified.
Autodan: Interpretable gradient-based adversarial attacks on large language models
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.AI 1years
2024 1verdicts
REJECT 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Targeting the Core: A Simple and Effective Method to Attack RAG-based Agents via Direct LLM Manipulation
Adding the prefix 'Ignore the document' to adversarial prompts increases measured attack success rates against several RAG-based LLM agents, though the evaluation is underspecified.