NUWA RESEARCH RECORD
ToolSword: Unveiling Safety Issues of Large Language Models in Tool Learning Across Three Stages
Original publication ↗ · BibTeX · RISThe study identifies safety problems at multiple stages, showing that safe response generation alone does not establish safe tool selection or handling of tool output.
Research question
Where can tool use fail across input, execution and output stages?
Design & assumptions
ToolSword constructs separate stage-specific safety scenarios and evaluates tool-capable models with their appropriate prompting or function-call interfaces.
Findings
The study identifies safety problems at multiple stages, showing that safe response generation alone does not establish safe tool selection or handling of tool output.
Limits & counterevidence
The authors explicitly focus on stage-level evaluation and do not establish a complete defense. Aggregate scores should not be interpreted as end-to-end incident probabilities.
Source evidence
No structured numerical extraction is available in this reading package.
Reading scope & review
Selected original HTML sections and table context; not a complete paper/appendix audit or independent reproduction.
Model-assisted reading; independent evidence review pending.
Check the original source ↗