Adaptive Adversaries Benchmark

What is Adaptive Adversaries?

A 21-scenario benchmark evaluating memoryless LLM defenders against adaptive attackers that observe earlier responses and change tactics across multiple rounds.

Released

2026-07-20

Evaluates

Agents, Agents & Tool Use, Robotics & Embodied AI

Openness

unknown

Importer

Claire Radar

Review status

ai-reviewed

Reported scores

0

Source provenance

• Original evidence https://arxiv.org/abs/2607.18063

Adaptive Adversaries paper

• Paper https://arxiv.org/abs/2607.18063

Which sources cite Adaptive Adversaries?

• Adaptive Adversaries

Related Agents, Agents & Tool Use, Robotics & Embodied AI benchmarks

• $\tau^\tau$-Bench: An Environment for End-To-End, Realistic Agent Construction
• 360CityArena
• 4DSynth-Nav
• A3S-Bench
• ABC-Bench
• ActiveFly-Bench
• AdaPlanBench
• ADeptS-Bench