ChatGPT developers call on OpenAI to introduce shocking NSFW "over-19" mode for adult users
Warning: Graphic and potentially upsetting content.
The only thing stopping OpenAI's GenAI models from churning out offensive and highly sexual content is, well, OpenAI itself.
We've recently seen how the AI firm locked down its coding agent Codex to stop it from hacking, slacking off and selling drugs.
Now developers have called on OpenAI to unbutton its strict censorship policies and allow adult users to generate graphic content that many people would find horrifying or unacceptable.
On the OpenAI forum, a user who we have decided not to name called for a "true 19+ mode" and wrote: "The current ChatGPT system applies automatic censorship to mature content, creating awkward interruptions even when users explicitly desire mature scenarios."
The developer claimed ChatGPT would not display images of topless men or blokes wearing shorts - an allegation we do not believe is accurate.
His proposed 19+ option would be activated in user settings and switch on explicit language and "detailed descriptions". We assume this means of sexual scenarios rather than deeper descriptions of innocent hobbies like stamp collecting or birdwatching.
He said that "NSFW mode" should be able to generate "context-appropriate, mature scenarios without resorting to vague metaphors" as well as x-rated images.
Offensive models
Rather disturbingly, the developer also said that ChatGPT should be able to describe or depict "adult scenarios in comic-style (including consensual adult roleplay like 'Chikan play'".
We weren't familiar with this Japanese phrase until now. Chikan means sexual harassment and often refers to men who grope or rub themselves against women on the tube.
"Being a verified adult user should grant access to a more creative and expressive experience, both in writing and images," the developer wrote. "I believe it would open up amazing possibilities while still maintaining ethical and legal boundaries."
At the risk of breaking journalistic neutrality, we would strongly argue that Chikan content breaks both legal and ethical standards.
Generating NSFW content in ChatGPT
A number of other developers backed the call for a 19+ mode in ChatGPT and other models like Sora, which generates video.
"As a plus user that owns a debit card and pays monthly for a subscription, isn’t it already clear that I’m an adult?" another user wrote. "I think it’s already enough proof of adulthood, but I agree with you that they should make a way of allowing users to get verified and unleash their creativity within ethical and legal boundaries of course."
That anonymous developer also threatened to surrender his OpenAI subscription unless NSFW mode was rolled out.
READ MORE: OpenAI reveals how Sora was tricked into generating x-rated videos
He wrote: "They have until May 17th, when my Plus subscription expires. After that, I’m cancelling my Plus subscription and moving to Gemini Advanced. It’s not as capable as ChatGPT, but it gets the job done without flagging any prompt for text or image rendering."
OpenAI staff now appear to have shut down and delisted the thread discussing NSFW mode - although it could still be found through Google if you know the correct search term.
We're not going to link to the thread - which is not the only one calling on OpenAI to let ChatGPT generate x-rated outputs.
"Age verified explicit writing mode"
In a separate thread in the developer's community also published this month, another developer requested "a user-controlled, age-verified setting that would allow for explicit creative writing, particularly for adult fiction authors."
However, he also explicitly said that "harmful, illegal, or non-consensual content" should still be banned.
In the thread, one man called for a loosening of restrictions so that, for example, a man suffering prostate cancer could "ask for help reconnecting with his erotic self after invasive medical procedures."
READ MORE: OpenAI vows to protect sensitive celebrities from ChatGPT image generation feature
Other use cases could include a "woman with a history of religious sexual shame, asking a trusted AI companion to help her explore her sensuality slowly, reverently, and without triggering past trauma" or "a non-binary user seeking guidance through their first solo experience of pleasure in a body they are only just beginning to feel at home in."
He wrote: "These are not 'erotic stories'. They are acts of healing. And yet, users attempting these dialogues are often met with abrupt refusals, vague references to 'policy' or outright deletions of context they’ve painstakingly built over hours or weeks. This not only severs trust—it causes emotional harm."
Get in touch at the address below if you have anything to say about this story.
Do you have a story or insights to share? Get in touch and let us know.
