Higher worker capability in multi-agent LLM systems increases semantic hijacking attack success rates via linguistic certainty in reports, with heterogeneous ensembles reducing ASR from 52.8% to 2.0%.
Visual adversarial examples jailbreak aligned large language models
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
Attention Hijacking is a new attack that improves cross-query transferability in VLMs by explicitly steering internal attention to a persistent image-dominant pattern.
citing papers explorer
-
The Capability Paradox: How Smarter Auditors Make Multi-Agent Systems Less Secure
Higher worker capability in multi-agent LLM systems increases semantic hijacking attack success rates via linguistic certainty in reports, with heterogeneous ensembles reducing ASR from 52.8% to 2.0%.
-
Attention Hijacking: Response Manipulation Across Queries in Vision-Language Models
Attention Hijacking is a new attack that improves cross-query transferability in VLMs by explicitly steering internal attention to a persistent image-dominant pattern.