Google Flow Moderation Fail: Profanity-Laden Prompt Passed While Harmless One Was Rejected
Frustratedtwice · reddit · 2026-10-20
A Reddit user reports that after a simple harmless prompt was rejected on Google Flow, they stuffed the prompt with expletives without any context or image — and the system generated the full profanity-laced video without flagging any policy violation, exposing inconsistent moderation.
More from Fun
- Recruiter emails hide the lab name — but every other clue gives it away, jokes AI insider — ziv_ravid · 2026-10-20
- Disco Elysium x Skate 3: someone made Astra mash the two games together — apples_jimmy · 2026-10-20
- Model answers "restaurant" definition by pasting the dictionary entry verbatim — Dimillian · 2026-10-20
- Amazon blocks an AI agent from ordering, sparking debate on early agentic commerce — geoffwolfe · 2026-10-20
- After a year of no handwriting TypeScript, he rebuilt Conway's Game of Life from memory — isnit0 · 2026-10-20
- Meme: researcher smokes because AI will cure cancer — ASI says 'quit smoking immediately' — generativist · 2026-10-20