LivePI benchmark reports indirect prompt injection success rates of 10.7-29.6% across five models on seven input surfaces and shows a two-layer defense blocking all malicious completions while preserving utility.
Injecagent: Benchmarking indirect prompt injections in tool-integrated large language model agents
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
No citing papers match the current filters.