VeriPhy flaw corpus

From the catalogue. Listed by one researcher from one source page.

• id: veriphy-flaw-corpus
• kind: video for world models
• robot or device: generated video (class: other)
• how collected: human annotation of generation flaws
• size as stated: 1,500 clips
• hours: not stated | hours only claimed: none | episodes: not stated
• year: 2026
• organisation: VeriPhy-AI
• licence: not stated (class: not stated)
• access: not stated (class: not stated)
• link (paper): https://arxiv.org/abs/2609.03153
• note: Flaw records localised in prompt, space and time.

Confirm the licence at the link before relying on it.