← back
arXivZiyan Jiang, Jingbo Yang, Jiabao Ji, Yujian Liu, Qiucheng Wu, Tommi Jaakkola, Yang Zhang, Shiyu ChangWed, Sep 30, 2026, 10:55 AM PDT
score 17.1

Benchmark Tests AI Agents at Finding Glitches in 3D Worlds

Original: WorldAuditBench: Interactive 3D World Auditing with Multimodal Agents

Source: arxiv.org ↗

Writing ELI5 summary…