Alright, who else is hitting this "unsafe content" brick wall with Sora? I've been through the wringer with AI video gen tools, and this one is giving me a special kind of headache.
I'm not trying to generate anything remotely controversial. We're talking simple, corporate-friendly scenes:
* A person walking into a modern office lobby.
* A product (a coffee mug) rotating on a table.
* A generic mountain landscape at sunrise.
Rejected. Every time. "Violates safety guidelines." What guidelines? The documentation is uselessly vague. It feels like they've cranked the sensitivity to 11, probably after some bad press, and now it's catching everything in its net.
My current theory is it's hallucinating "unsafe" elements. Maybe the texture on the office floor looks like something else? Maybe the mountain shadow is "suspicious"? Without any actionable feedback, I'm just shooting in the dark.
So, has anyone actually cracked the code on this?
* **Are there specific keywords in the prompt that are instant triggers?** (I've avoided all the obvious ones).
* **Does using certain styles (e.g., "photorealistic" vs. "animated") have a higher chance of tripping the filter?**
* **Has anyone gotten a rejection overturned by support, and if so, what was the actual reason?**
This is a major workflow killer. I can deal with weird artifacts, but this opaque censorship is something else.
been there, migrated that