ChatGPT imagegen guardrails loosening but inconsistently, user tests show
flowersslop · x · 2026-11-20
User flowersslop reports that ChatGPT's image generation has become noticeably less restrictive lately — but inconsistently. Completely harmless prompts still get blocked, while other requests randomly pass through, including an image with exposed underboob and nipples (shown in the attached image). The observation suggests OpenAI's image model safety guardrails are in an unstable, unevenly-loosened state.
Related event: Users Report Loosened and Inconsistent ChatGPT Image Generation Guardrails(2 posts)→
More from Models
- Google ends free Gemini Pro student subscriptions worldwide as model matures — Angaisb_ · 2026-11-20
- Perplexity's Decider V1.3 tops RollerCoaster Tycoon decision benchmark while free on OpenRouter — antoine_chaffin · 2026-11-20
- GPT-6.2 System Card Adds First Model Spec Compliance Section With 738 Tests — sjgadler · 2026-11-20
- LoRA adapter lifts Qwen3.8-27B to 63% on SWE-Bench Pro, 60% fewer tool calls — ben_burtenshaw · 2026-11-20
- GPT Pro veterans are cancelling over 6.2 quality and new Chat quotas — lxfater · 2026-11-20
- Model UX: why Google's writing style keeps users coming back — flavioAd · 2026-11-20