SafeVid Benchmark

What is SafeVid?

SafeVid is a 20K-video benchmark for detecting AI-generated videos across 10 social risk categories. It tests physical plausibility, structural consistency…

Released

2026-07-03

Evaluates

Agents, Reasoning, Robotics & Embodied AI, Vision & 3D

Openness

unknown

Importer

Claire Radar

Review status

ai-reviewed

Reported scores

0

Source provenance

• Original evidence https://arxiv.org/abs/2607.03069

SafeVid paper and code

• Paper https://arxiv.org/abs/2607.03069
• Code https://github.com/williamw99/SafeGuard

Which sources cite SafeVid?

• SafeVid

Related Agents, Reasoning, Robotics & Embodied AI, Vision & 3D benchmarks

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