AICompanionBench

What is AICompanionBench?

AICompanionBench provides a dataset of 2,123 human-AI companion conversations with safety risk annotations to evaluate LLMs-as-judges for detecting unsafe…

Released

2026-06-03

Evaluates

Robotics & Embodied AI, Safety, Safety & Trustworthiness

Openness

unknown

Importer

Claire Radar

Review status

unreviewed

Reported scores

0

Source provenance

• Original evidence https://arxiv.org/abs/2606.04867

AICompanionBench paper and code

• Paper https://arxiv.org/abs/2606.04867
• Code https://github.com/anonymousresearcher2026/AICompanionBench/blob/main/AICompanionBench.xlsx

Which sources cite AICompanionBench?

• AICompanionBench

Related Robotics & Embodied AI, Safety, Safety & Trustworthiness benchmarks

• 360CityArena
• 4DSynth-Nav
• ABC-Bench
• ActBench
• ActiveFly-Bench
• Adaptive Adversaries
• AeroGround
• Agent Planning Benchmark