Roblox downloaded 145GB models from us (including adult content) without asking
- ID
- 40022
- Status
- summarized
- Published
- 27 Oct 2026, 6:27 PM
- Fetched
- 27 Oct 2026, 11:50 PM
- Provider
- Hacker News
- Category
- dev-community
- Original URL
- https://open3dlab.com/news/roblox-downloaded-145-gb-of-models-from-our-sites/
- Source URL
- https://hnrss.org/best
Summary
- Score
- 7.5
- Created
- 27 Oct 2026, 11:50 PM
- Tags
- Audience
- developersai_ml_learnerssaas_startup_founders
What happened
On 15–16 October 2026, a crawler identifying itself as trellis2-data-sourcing/1.0 (3D foundation model research) and using a single AWS address read 18,926 project pages across Open3DLab, SFMLab, and SmutBase, then downloaded 1,027 complete files from 712 projects totaling 145.5 GB. The affected sites say 345 of those projects (62.6 GB) were adult content and 423 were licensed non-commercial-only; they blocked the crawler on 18 October, received no reply to their email, and saw blocked retries on 19–20 October.
Why it matters
If you host user-generated 3D assets or downloadable content, this incident shows robots.txt and ToS were retrieved and ignored, and the main cap on volume was free-download speed and concurrency limits. Decide whether downloads need authentication, stricter per-IP/rate limits, licence metadata, and alerts for catalogue-wide page reads, and treat scraped 3D datasets as a provenance and licensing risk if you train or ship models.
Discussion angle
What would actually have stopped this crawler sooner: auth-gated downloads, tighter rate limits, or legal/licence enforcement—and would those controls break legitimate users?