Sora, Your content policy is killing creativity

[Mark_Almencion]

I just want to say how frustrating it is to use Sora when you’re trying to build something original — especially when all your generations get blocked by vague content policy flags.

I’ve been working on a fictional character inspired by anime, turning him into a stylized, hyper-realistic version. All assets are AI-generated — not real people — and yet, almost every edit gets blocked under “review remix.”

This isn’t about making anything harmful. This is about creating characters, evolving their look, testing lighting, mood, and expression. But Sora treats basic adjustments — like hairstyle tweaks, visual upgrades, or tone changes — like red flags.

Why are you punishing creativity?

I’m paying for a tool that constantly tells me “No,” while giving zero clarity on what I can do. Either explain your limits better — or trust your users more.

[jandousek]

Sora has no idea of the context. She sees an inappropriate combination of words or just some words and sends you to hell. Me, when I want to create something sensual, like an android standing in the rain (NOT NAKED) and covering her lap with her hand for my story, I have to perform incredible verbal haiku while others jailbreaking it and drawing vaginas for fun.

Yes, I’m forced to jailbreak it too in some sense.

So yes, the restrictions are tighter, but I’m afraid they are necessary. People can’t be trusted. And they certainly can’t be trusted by a company trying to establish some credibility.

I recommend trying the uncensored models.

[Mark_Almencion]

but what they are doing is overkill.. like i feed a prompt for hyper realistic upgrade of an anime character, (wholsome and clean) and sora submit its input, if i wanna request to change the age to 3 years younger, it non compliance to their content policy?. its insanity.

[jandousek]

Looks like a bug. And ridiculous one. But you don’t know how I understand. I’m tired of making up nonsensical prompts just to get Sora to draw something normal. I’m already thinking about getting a new PC and starting my own model.

[jandousek]

They must have done some kind of update again. Things that I could normally draw, even if I had to describe them in a convoluted way, no longer work for me. I can’t even say that the character has an absolutely flat chest (my character just doesn’t have breasts, period).

What do I write about and what do I draw? Kittens? Funny, in a sense.

Just draw this took me lot of red walls. Sensual? Yes. Not a porn. Not in the slightest. I still had to perform crazy haiku to make Sora understand that I want her breasts covered. It’s beautiful and it’s hard, when Sora doesn’t understand the context, to create it.

[massimo.cusi]

Frankly, it’s become nothing short of cretin. In the desperate and futile attempt to avoid backlashing from extremists and play virtue signaling to the extreme, the paying user is rendered unable to generate anything beautiful and attractive.
Look, no one cares about the ordinary, the inclusivity mantra has never had any point. Our imaginations roam free, they create world of beauty and ideal perfection. Sensuality, beauty and whatever some extremists would erroneosly call names are what make us humans.
The paying customer is being met with ridiculous, completely ideological and repressive policies that prevent anything worth seeing being created on this and many other platforms.

Meanwhile, the public section of generated “art” shows scores upon scores of the most extemely pointless, stupid, not to mention irrelevant junk that doesn’t get blocked by the policies. Loads upon loads of cretin videos of car-driving cats, parachuting pigs, and banal memes on irrelevant social commentaries by laymen who have nothing else to say but to repeat the same trite propaganda with pictures and signs.
Sora is a place of nonsensical censorship where nothing meaningful will be ever able to pass through the cross fire of “policies”. Nothing that does disturb the viewer, that does provoke thought, can be produced on this platform.
There are thousands of irrelevant, dull, completely insignificant blobs of subpar productions that pretend to mean something when they don’t, but nothing that will be recalled past the next 15 minutes by any viewer.

If this is what openai has put billions of dollars on and spent millions of hours in training, it is but a huge waste of time and wealth to produce nothing.

Practically everything that gets past the censorship on this model and any other wasn’t worth creating in the first place.

One wonders why we keep paying for a service that’s not even delivering half of what it could and what little it does is making mankind worse.

[carlos.portillo]

Unpopular opinion here: If you don’t trust your users, then get out of business. You can’t expect people to pay for a tool you are not allowing them to use because you don’t trust them.

Is like Ford sells you a car but doesn’t give you the keys because they don’t trust you would drive safely.

And here is the thing: they are mostly punishing us because of stock price. Yes, that’s the whole reason. If OpenAI was not into making money like most SD models out there, they wouldn’t care. They don’t do it because of “our security” but because of their “financial security” and that just sucks big time.

[jandousek]

That’s really not a way to force it. The rules are clear. Yelling won’t change anything.

[saevusladon]

Yep, I pretty much gave up on trying to make anything meaningful in Sora, the recent update patch back in May tightened up against paying users because “minors” found a way to cheat the system and get NSFW content using medical terms.

I can’t even use chatGPT for anything anymore with image creation, they’ve tightened the leash so hard that mostly only G-rated content is safe to make.

It was never about “user safety” its about protecting company assets. (Aren’t you supposed to be an NPO )

This is a lie.

I hate being lied too so blatantly.

This is the same nonsense the left wing did to push their censorship ideology back during the Biden Administration. Restricting content or removing it under the guise of “Safety”.

It’s nonsense and a blatant bait/switch tactic to lure in subscribers, then jerk the chain.

[carlos.portillo]

I’m not trying to force anything. And no, the rules are not clear. That’s actually the problem. Here is a link with the guidelines:

https://openai.com/policies/creating-images-and-videos-in-line-with-our-policies/?utm_source=chatgpt.com

I don’t see how prompting sora to change pants for shorts, or a sweater for a tshirt is a violation of these policies. I have been flagged even for asking for a hairstyle change has been flagged. I once put a picture of me and my girlfriend standing in the street and asked to change it to a beach environment and it was flagged as inappropriate content.

The written policies are clear and specific. The problem is OpenAi is flagging beyond the specifics of their pwn content policies, in an attempt to be “politically correct”, just to please minorities with crystal personalities who will still find a way to be offended anyway. And that is hurting creativity and limiting us, paying users, a lot. I don’t really see why they have invested so much money just to give us a cat meme generator. It’s idiotic.

[saevusladon]

[morpheus23x]

Sora is Amazing! But I’ll never buy it, because I’m never going to let anyone tell me what I can or cannot make.

[MavStevens]

Please add better transparency to why something may violate the policies. I made a video no problem, and tried adjusting the prompt to remove dialogue. Changed nothing about the prompt, just added a line saying they don’t speak about it. BAM content policies. It’s infuriating and really does completely murder creativity

[TiKkO_Rome]

Anime fans are fans of the artist. Everyone knows what AI can and the “wow” factor has worn off. So, draw it yourself or collaborate with an artist who likes your ideas. If you wanna be thought of as a creative person, you might try starting by actually being creative. Create art, instead of more slop.

[fusiondesigner]

They seem to have all these content policy police and copyright vagrants working there atm. Just nimbies making policies so someone doesn’t get their feelings hurt.

[roost_the_cockatiel]

Same thing with Bing image creator, I tried making a picture of a male J-pop idol in purple spandex and they say that it’s unsafe, same thing with Grok imagine, I’m trying to make an autistic fourteen year old Japanese boy OC and they keep moderating it, same with all the other projects I’ve been doing

[saevusladon]

Looks like there was another major update last night for both Sora and the GPT platform for image creation.

Image quality changed dramatically for the exact same prompts. On the first picture you see the older Sora Quality. (Same on GPT platform btw ) and the “New” quality with the exact same prompts.

Ingoring the artifact errors, you can clearly see a huge mismatch between image quality.

It feels so watered down.

From what I gather from some of my tests:

I originally thought I was pushed into the “New Safer” “For minors” only platform because of being erroneously flagged as potentually under age , [I did not recieve any banners, prompts, or emails about it ] but further tests seemed to debunk that hypothesis by taking prompts from other users on the SORA posting board and getting the exact same results with their prompts.

So what I can gather is prompts are now being interpretted differently and Sora/GPT may be applying its own stylishic logic based on what you prompt for even if you don’t request it.

As with any “Update” Sora/GPT are now even more strict and aggressive with blocking content.

Seriously, I don’t publish to the public domain, I just create stuff visually for my own purposes to go along with the novels I write. There is literally no reason to be this hyper aggressive at blocking content that isn’t even meant to be sexualized, just athetically and contextually accurate.

This is the opposite of what we’ve been asking for Open AI, and what you’ve been promising us with “Treat adults like adults”

I now feel more like a kid than I ever did using your services as a paying subscriber.

[TiKkO_Rome]

If its about creativity, why aren’t you using real humans? Artists, actors, sound engineers, whatever you need there are literally millions of humans as passionate about that aspect of what you’re trying to do as you are about whatever creative contribution you’re actually making.

Dont get me wrong, I use AI too. It takes me hours or even days to search my code for the cause of bugs in the software while simultaneously cross-referencing all relevant manuals, guides, etc., whereas ChatGPT can do it in minutes. I use it to speed up my work, which is exactly what technology is supposed to do. I dont subcontract my creative work to AI and then claim to be creative. If I’m working on a song and find I need a guitar part, I don’t ask AI to create it. I post in a guitar group on social media and have 100 skilled guitarists offering to play the riffs for cheap and even free the same day. Thats not MY creativity on that guitar part, but it is another person which adds value to my song because nobody is impressed by nor connects with AI. Faux-art (by definition, art cant be made by AI) is a NOVELTY. People look at it and go “Wow! Its crazy that AI can create that” and the next day nobody gives a shit because its a tool, not an organic being who people identify with. Art is impressive because people recognize what it took to develop those skills, their’s admiration, envy, respect, etc.

Mariah Carey is one of the most gifted singers of all time. Her range is phenomenal, but she cant go as high as a keyboard/synthesizer. Nonetheless she’s regarded for hitting that note and nobody bats an eye at the keyboard because its just a tool. Eminem is a legendary lyricist renowned for his syllabic structures. AI has yet to be able to match that, but once it does, people will say “Wow, I didn’t realize AI had come so far” and then they’ll never give a shit again because its just a tool.

Milli Vanilli was a HUGE pop sensation in the 1980s. Then, people found out it wasn’t actually them singing and people stopped listening to them overnight and that’s with there actually being a human performing the art. The real singer would’ve become a star overnight, but labels didn’t introduce him to the world in time for people to learn who he was and connect with him and , thus, reconnect with the music. Without a person to connect the art to, people stopped listening.

Ashley Simpson was a rising pop star until a mishap on SNL revealed she was lip-syncing her song, which gave credibility to the rumor that auto-tune was responsible for her singing ability and her career crashed overnight.

Comic book fans talk about the artists and writers, film buffs talk about directors, actors, writers… The point is that, if there’s actually something creative you bring to the table, don’t void your gifts by relying on AI to do it. If, like most people in your position, do NOT have any talents to speak of, that’s okay. The fact that most people cant is what makes the people who can special