SchemeArena Benchmark

What is SchemeArena?

Evaluates scheming in LLM agents across 400 factorized scenarios, with a monitor that scores covert misaligned behavior along five rubric dimensions.

Released

2026-09-08

Evaluates

General AI, Safety, Safety & Trustworthiness

Openness

unknown

Importer

Claire Radar

Review status

ai-reviewed

Reported scores

0

Source provenance

• Original evidence https://arxiv.org/abs/2609.08126

SchemeArena paper and code

• Paper https://arxiv.org/abs/2609.08126
• Code https://github.com/launchnlp/SchemeArena

Which sources cite SchemeArena?

• SchemeArena

Related General AI, Safety, Safety & Trustworthiness benchmarks

• $\tau^\tau$-Bench: An Environment for End-To-End, Realistic Agent Construction
• 3DCodeBench
• 5-Dialects-BN
• A3S-Bench
• AARRI-Bench
• ActBench
• ActiveSciBench
• ActiveVision