Safety of Latent Communication in Multi-Agent Systems (Q7361999)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
No description defined
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | Safety of Latent Communication in Multi-Agent Systems |
No description defined |
Statements
Safety of Latent Communication in Multi-Agent Systems (English)
0 references
30 September 2026
0 references
cs.AI
0 references
cs.LG
0 references
cs.MA
0 references
Muhammad Huzaifa
0 references
Sina Mavali
0 references
Thorsten Eisenhofer
0 references
We study latent communication in multi-agent systems and find that benignly trained links can increase harmful compliance compared to text-based communication. A reinforcement-learning attack that rewards both harmless response accuracy and harmful compliance further amplifies this effect, raising the mean harmful-compliance score from 27.9 to 76.9 across three topologies and four safety benchmarks.
0 references
Latent Communication
0 references
Safety Alignment
0 references
Multi-Agent Systems
0 references
