AuthMem-Bench
Explore score history ↗AuthMem-Bench evaluates authority collapse in persistent memory for LLM agents. It uses a paired benchmark holding claims and tasks fixed while varying source authority, measuring write-time collapse, authorization errors, and automatic authority preservation.
- Released
- 2026-08-03
- Institutions
- Tsinghua University · East China Normal University
- Evaluates
- Agents, General AI, Language & Knowledge, Safety, Self-Evolution
- Openness
- unknown
- Importer
- Claire Radar
- Review status
- unreviewed
AuthMem-Bench paper
No reported scores are on record for this benchmark yet.