A new data synthesis framework improves LLM agents' refusal of harmful multi-step tool-use requests with minimal impact on helpfulness.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
AgentAlign: Navigating Safety Alignment in the Shift from Informative to Agentic Large Language Models
A new data synthesis framework improves LLM agents' refusal of harmful multi-step tool-use requests with minimal impact on helpfulness.