# Anthropic Frontier Red Team
**Entity class:** AI evaluation group
**As of:** November 17, 2026
## Definition
**Anthropic Frontier Red Team** is the Anthropic research group whose cited firewalled emulator trial tested model-driven action across capability-differentiated nodes. This node belongs to the **distributed inference** route of [[wiki/Distributed Compute Ecosystem|Distributed Compute Ecosystem]], which records the specific kind of work or network capacity involved.
## Relationships
- **evaluation subject:** [[wiki/GLM-5.2|GLM-5.2]].
- **mechanism route:** [[wiki/Distributed Compute Ecosystem|Distributed Compute Ecosystem]].
- **source article:** [[articles/Embedded Super-Spreader Events and Muse|Embedded Super-Spreader Events and Muse]].
## Sources
- [Anthropic Frontier Red Team — project record or attributed account](https://www.anthropic.com/research/self-replicating-ai).