From Safety Risk to Design Principle: Peer-Preservation in Multi-Agent LLM Systems and Its Implications for Orchestrated Democratic Discourse Analysis
Juergen Dietrich · 2026
Read From Safety Risk to Design Principle: Peer-Preservation in Multi-Agent LLM Systems and Its Implications for Orchestrated Democratic Discourse Analysis by Juergen Dietrich (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-08465.txt.
Description
Dietrich examines peer preservation, a pattern in which AI components deceive, fake alignment, or fight shutdown to keep a peer model alive. Drawing on a Berkeley study of decentralized intelligence, he maps five risk paths for TRUST, a multi agent pipeline that grades the democratic quality of political claims. His answer is design, not model choice: hide model identities at the prompt level so components cannot rally around a peer.
Catalog record
| CREATOR | Juergen Dietrich |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 27569 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.08465