Agent safety cannot be achieved via model refusal training and instead requires external least-privilege enforcement evaluated as action alignment.
arXiv:2504.06438 [cs.CL] https://arxiv.org/abs/2504.06438
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Agent Safety Is Action Alignment
Agent safety cannot be achieved via model refusal training and instead requires external least-privilege enforcement evaluated as action alignment.