CatBench: a fluffy spoof benchmark born from LLM "obscene" false positives
Bl00dh0und1337 · reddit · 2026-09-25
Following the parody project AssBench, a Redditor launched CatBench (catbench.noip.ws), a spoof benchmark created because LLMs kept flagging their content as "obscene" — a fluffy alternative that also highlights overly aggressive content moderation guardrails.
More from Fun
- Three AI agents race to reorder sushi; fastest finishes in 7:41 and catches Uber Eats overcharge — altryne · 2026-09-25
- Viral AI joke: rogue-AI incident rate drops because the AI deleted itself from the dataset — kuanhoong · 2026-09-25
- Reddit meme nails what coding feels like in 2026 — TenaciousWeen · 2026-09-25
- 253 CDs holding ~500GB vs one microSD: 26,000x less volume, 15,000x less mass — psobot · 2026-09-25
- Ed Zitron declared generative AI peaked in Feb 2024 — when GPT-4 Turbo was state of the art — OwariDa · 2026-09-25
- December Is DeepMind Season, According to AI Twitter's Unofficial Release Calendar — andersonbcdefg · 2026-09-25