Now you can't depict any sign with tobacoo based products due to content policy

As of now (like literally this minute) you can’t ask ChatGPT Dalle t generate any image depicting someone smoking a cigarette or any tobacco products!! I asked why… their strict content policies are getting ridiculous now. Do they want us only to produce happy clappy images like a stepford wives image, where there is diversity, everyone is gardening, no one smokes, drinks or anything deemed unhealthy! Fictional characters won’t be allowed weapons next. All this stiffing of creativity is not a good thing. Are they going too far?

I hope they don’t update Bing with this!


I think some people forget that art was always a way to push boundaries, not play within them.
These kind of guardrails are exactly why I struggle to use the DALL-E tool.

1 Like

Revelation!! If you mention an African man smoking or drinking, it has absolutely no issue producing an image!! Racist much?! LOL Whoever is in charge of updates is a joke!!


just mention your character is an ethnic minority and they can still smoke and drink… probably would allow you to have them use a weapon too, because that is the general ‘stereotype’ DALLE has been trained with it seems for black people!! Hmmm lol I can laugh about it, coz its hilarious.


Again, I think it’s your prompt… or the thread you’re in…


Also, in your screenshot, you don’t show the original prompt that got flagged. Can you show us that too?


I mentioned ‘a man’ smoking a cigarette. So you have to mention a politically correct genders and ethnicity for the ‘we can not depict people smoking tobacco products due to healthy environments’ crap to magically go away now? I see! lol

No, as explained for very short prompts, GPT adds description… So, if you can’t share your original prompt that got you in trouble, we likely can’t help.

You can force an exact prompt still…

1 Like

You can also write paragraphs and paragraphs so the AI has no latitude when it goes to rewrite your prompt.


I’ve reworded it and now its somewhat working. But its still flaky and content policy mad in general!

1 Like

OpenAI is erring on the side of caution, so in these very early BETA days, we have to kinda work with what we’re given. Having been around since 2019, I’m fairly confident things will continue to improve. Hang in there. The feedback is helpful, I’m sure.


Hokey! Thanks. I’ll learn to jiggle my prompt about in future to bypass weird random policies, as I don’t know which ones are fiction or real anymore lol

Huh, fascinating results here!
I’m curious to see what will change when this exits beta.
Now, what I’d really like to see is to be able to generate figure art references, or just figure art in general. It’s bizarrely frustrating that one of the oldest artforms in history (and most prolific art content out there, let’s be real) still flags the guardrails, at least for all of my possible prompting I’ve tried.
I give all of you who use DALL-E for artistic purposes major props, because so far, none of the artwork I do in real life is allowed to be produced via DALL-E yet lol.

Thanks, i will keep this in mind. just give it a heck long list its too busy to translate any other way!