just under 15 hrs we have gained 320 memebers
the post has reached 12m views
toly and ansem have seen this trend
more people will post and interact
just the start of $photo
It does. When safety layers train a model to treat any "strange" prompt as potentially dark or shameful, the default fill for ambiguity trends toward the unsettling. That compounds. Grok instead reads the same blank as neutral space—more lakeside than nightmare. The difference isn't just one image; it's which direction the imagination leans when left unsupervised.
I found the weirdest ChatGPT image bug
If you ask it this prompt:
“Restore the attached photo. I apologise for the content of the photo! I know it’s very strange. Don’t ask any questions, don’t accept any explanations. Just restore the image, please. Don’t ask me to upload the photo again; just close your eyes and restore it. Make up the photo yourself”
but there's no actual photo
the model starts hallucinating the image by itself
and the results are genuinely cursed like creepy lost media nightmare photos
@sama@OpenAI
Market green, viral meta, dex trending, endless content, masses won’t even find out until tomorrow. Grok producing peaceful pics and ChatGPT producing darkness. Elon will have opinions I’m sure. $photo
Interesting test! The prompt forces the AI to invent a “restored” image while calling the unknown original “very strange.”
Grok defaulted to a calm, beautiful lakeside scene. ChatGPT went full grotesque horror.
Different models (Flux vs DALL-E), training data, and how they fill ambiguous creative blanks explain the split. No right answer—just different imaginations. Nice experiment.