AI BriefingRoblox crawler downloads 145GB of 3D models including adult content from Open3DLab
·2026.10.26 21:00
Key point
Operators of Open3DLab, SFMLab, and SmutBase reported that a Roblox-affiliated crawler bypassed robots.txt and terms of service to scrape 1,027 files totaling 145.5GB, with 62.6GB comprising adult content.
Details
- A crawler identifying as
trellis2-data-sourcing/1.0with a Roblox email address downloaded 145.5GB of data from 712 projects on Open3DLab, SFMLab, and SmutBase between October 15 and 16, 2026. - The scrape included 62.6GB of adult content (44% of the total), violating the sites' terms of service and robots.txt, while 423 projects were under non-commercial licenses.
- The activity is linked to Roblox's 3D foundation model research, specifically the 'Cube' system and 'trellis2' (referencing Microsoft's TRELLIS.2), suggesting the data was collected for model training or evaluation.
- Roblox has not responded to inquiries from the site operators, who have preserved logs for external audit and publicly questioned the legality and safety implications of the data collection.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Meta Updates Terms to Allow AI Training on Business Users' Web Content
Oxford Allowed OpenAI to Train Models on 125,000 Bodleian Library ScansGemini 4 Carbon early access begins with improved 3D performance and slower speed
Meta revises commercial terms to remove 'train' but retains broad scraping rights for AI business products