Independent 3D modeling community Open3DLab warned that a web crawler linked to the gaming platform Roblox mapped nearly every project on its website, along with sister sites SFMLab and SmutBase, downloading a total of 145.5GB of data. The group reported that the crawler sourced complete copies of 1,027 files from 712 projects, which Open3DLab believes were taken for AI training purposes. Open3DLab staff first noticed the crawler making a large number of requests to its sites and download infrastructure on October 18.
The crawler utilized the user agent trellis2-data-sourcing/1.0 and originated from a single Amazon AWS address. Open3DLab stated that the crawler failed to respect the site's robots.txt file or Terms of Service, which explicitly forbid scraping, and also ignored creator licenses that prohibit non-commercial use. Over a span of 34 hours, the crawler read practically the entire catalogue of 18,926 projects, executing 26,429 requests before suspending operations.
The downloaded assets included 364 files, totaling 54.8GB, from 224 projects on SmutBase, and an additional 345 projects of adult-marked SFMLab projects, amounting to 62.6GB. Open3DLab noted that the crawler's main limiting factor appeared to be its use of free download servers that capped speed and concurrent downloads. The scraping activity was detected on October 18 and blocked by the following morning, though further download attempts were detected on October 19, which continued despite blocking. A final attempt was made on October 20, with requests tapering, which Open3DLab suggested was
