Hidden in Plain Sight: Benchmarking Agent Safety Against Decomposition Attacks with DECOMPBENCH
Factual evidence
What the source reports
Researchers introduced DECOMPBENCH to evaluate LLM agent safety against decomposition attacks that split harmful tasks into benign subtasks.
Inspect the evidence
- Inclusion basis
- Enterprise AI
- Publisher and source type
- arXiv cs.LG — Machine Learning · RESEARCH
- Published by source
- 8 October 2026
- Collected by OneBench
- 9 Oct 2026, 03:02 UK
Stored source excerpt
arXiv:2606.13994v2 Announce Type: replace-cross Abstract: LLM-based Agents are becoming increasingly capable and widely deployed, creating growing incentives for adversarial misuse in the real-world. A key…
Short excerpt from the collected text, not the full source. Use the source link to read it in context.
The factual summary is a OneBench synthesis, not a quotation or independent verification. Collection time is not publication time. Open the source for its full context; related reporting can share the same underlying announcement.
OneBench interpretation
Institutional assessment
So what
Multi-step agent orchestrations in financial workflows face novel safety bypasses when harmful intents are split into benign subtasks.
Do what
Review agentic architecture guardrails with the model security team to ensure subtask safety checks evaluate aggregate intent.