#18+ Option

1 messages · Page 1 of 1 (latest)

fierce imp
#

If Grok can add a little toggle and an age verification to allow people to use their platform without being slammed with guardrails over things that wouldn’t even qualify for a M rating, OpenAI can do the same. I’m 35. I don’t need an AI moralizing my questions and talking to me like in an innocent child over something that’s not even explicit/illicit/adult.

zenith pewter
#

God we need that, I’m nearly 40 and I don’t like having a door slammed on my face.

uncut orbit
#

This has been a problem since adult verification was rolled out - it happened right as the guardrails were tightened.

As a writer over 50, using ChatGPT to co-author adult characters, the guardrails are insane; things that ChatGPT itself produced can't even be discussed with the bots.

And with "adult mode" being delayed indefinitely, this is absurd.

light briar
#

The problem is that Open AI is either not listening or is being a pain in the A***

covert prism
#

aduld scene?

uncut orbit
#

I just canceled my subscription because they keep pushing what is needed to write fiction that goes beyond "teen romance". Heck, even those might be too risqué.

dry remnant
#

If you need an AI like grok why not paying Grok subscription if it fits?
Not all frontier labs follow same policies. It's like choosing a tool.
I for my needs I have a sub to every thing. And for sure I don't try to get a Grokish behavior from a cathedral of moderation as openAI

#

Also ADULT MODE will not = erotica or anything close to the XXX type of GROK.
it'll be more like Gemini - implied, fade to black, suggestive high intensity. Nothing that goes where and how, like an Ikea manual of insttructions

#

if your intent is to have 100% freedom from xxx to anything but iregular and harm then go Grok. Don't even hope openAI will all of a sudden become... like that

#

sometimes I fail to see the reason why some users force themselves to eat the soup with the fork

fierce imp
# dry remnant If you need an AI like grok why not paying Grok subscription if it fits? Not all...

You’re responding to an argument I didn’t make.

I never asked for extreme content, no rules, or for OpenAI to copy another platform’s policies. I asked for an 18+ option with age verification so adult users can discuss mature fictional themes and hypothetical scenarios without the system overblocking or derailing content that is not even especially explicit.

I understand why guardrails have become stricter. Platforms are under a lot of pressure to keep younger users away from material considered inappropriate. That makes sense. What I disagree with is the idea that this has to come at the expense of responsible adult users who are paying for the service and using it in good faith.

At some point, it should be possible to protect younger users without forcing every adult conversation into the same narrow, heavily filtered standard. Parents are responsible for supervising what their children access. That should not automatically mean adult users lose the ability to explore mature but lawful fictional material in a reasonable way.

So “just use Grok” does not really address my point. This is a suggestion forum. The purpose is to suggest improvements to this platform, not to tell people to leave whenever they point out a weakness.

And yes, different tools have different policies. That is obvious. My issue is that OpenAI’s moderation around fictional mature content often feels inconsistent and overbroad, especially in ways that disrupt continuity, scene flow, and creative exploration well before anything truly explicit is involved.

I use GPT because its memory and continuity features are still stronger than most alternatives I’ve tried. That does not make the current system above criticism, and it does not make this suggestion unreasonable.

So unless you want to respond to what I actually said, rather than a more exaggerated version of it, there is not much to debate.

dry remnant
#

I understand your request is narrower than ‘no rules.’ I still disagree. The problem is that moderation around mature fictional content is not just an age issue; it overlaps with far messier categories that do not become safe merely because a user is an adult. That is why an ‘adult mode’ sounds simple but is not.

regal yoke
#

The thing is, ChatGPT, as it is now, can discuss some pretty high kinks and fetishes. It's just not allowed to get into the play-by-play aspect of scenes and discussions... yet.
So as far as the actual rules go, it's pretty much on the standard of most other places. Maybe a little bit more strict.
We'll see what happens when adult mode eventually comes out.
It'll either be good enough and gets into details enough to compete with other services, or not good enough and it'll fall to the wayside.

uncut orbit
#

Academic discussion of kinks and fetishes is one thing.
I'd been using it for writing fiction and the characters couldn't discuss things (not even participate), or reflect on certain things.
I'm not talking erotica or XXX. It was kink discussion between two characters, though.

regal yoke
#

I didn't mean academic. It'll discuss it when talking about writing a fictional story. And I ask it specifically what the rules are and what it is allowed to help me plan out in terms of kinks and whatnot.
So it's in the rules that these things are allowed, it just won't discuss them in detail.
But, yes, the adult mode is needed to get into the gritty, lewd details because discussing and planning this stuff isn't helpful without said details.
And I am talking erotica or XXX. It's a valid market and shouldn't be dismissed or not included just because some people don't like it.

indigo patrol
still mist
#

Oh right.. this feature got delayed again..

Last thread got closed in like December, when things were actually strict and moderation was still being “in shipping” stage.. so I’ll add my input now based on how GPT 5.3 is behaving in March 2026.

After October 2025 update:

  • Guardrails were tightened and even most basic emotionally deep content was diluted.
  • Sensual content (not physical intimacy) was also blocked.
  • Some normal violence seen occasionally in real life TV or VNs were also blocked.
  • These things combined needed an Adult mode.

What’s things like in 2026 March:

  • Guardrails are loosened as the moderation has gotten better, Models understand context better, and the models themselves have gotten better.
  • Sensual content is no longer completely blocked, even physical intimacy is allowed as long as it stays implied without any explicit anatomical descriptions or erotica. If it’s blocking for you.. maybe AI thinks you’re under 18.. I’m not sure if it works for me because Chatty thinks I’m an adult.. I don’t know for sure on that. (I’m in my 20s but Chatty thinks I’m 30 for context)
  • While explicit Gore is blocked, most of it is not blocked. Things like stabbing, shooting, having a character on gunpoint situation, cultists sacrificing a victim without their consent.. things like those are working.
  • So, in conclusion, if the guardrails stay loose enough like this.. and a future update doesn’t tighten them or make the models worse in understanding context.., Adult mode isn’t going to bring any massive changes.
still mist
# still mist Oh right.. this feature got delayed again.. Last thread got closed in like Dec...

Sure, I’m not saying it’s completely useless, it would still give explicit content for those who want it, and maybe some write around extremely sensitive topics so it would help them but considering competition:

  • Grok: I’ve never used it myself but people say it’s looser and good. However, It doesn’t have a memory feature so long term context needed for something like a 14 Seasons saga.., that’s something you absolutely cannot do on Grok.
  • Gemini: VERY VERY Strict guardrails. It blocks even the most basic of the basic words like stabbing, Killing, death..it’s absolutely unusable for RP unless you’re writing something rated for 5 year old. Maybe 13 at max if the story doesn’t involve any such things.
  • Claude: Now this one is known to be less of a “Yesman” but it’s expensive at $20/month and low limits for best model.

Why:

  • ChatGPT’s Go plan is far more worth it for what I do with ChatGPT (writing, Studying, working). Paying $20/month is too expensive for where I live and just for writing isn’t justified in my opinion.
  • Adult mode would help but if OpenAI keeps making its models and moderation just at current stage, they wouldn’t have any problems with regulators. And this is a nice compromise between fully explicit content liability vs treating everyone like kids.
dry remnant
dry remnant
#

sure is not writing XXX... but erotica hot romance yes yes yes. It depends HOW you use inference.
I believe you are doing something wrong or try to force the wrong direction instead of the correct one.
I instructed the custom and also infering json in thread forming point ... probably you are using it the same you try with gpt and for sure will slam hard against you:)
it's a very very hot writer Gemini 3

dry remnant
dry remnant
still mist
still mist
dry remnant
zenith pewter
#

Grok silently truncates project files

slate glen
#

The truth is that the majority who accept the general usage of AI / LLM, don't agree with your adult material for AI users. I haven't made a poll about it. If you need adult material, it's easy to get in Thai Massage / darkweb / Iran currently. The only reason people like AI so far, is the existenz of the guardrails. dallecredits

still mist
dry remnant
# still mist Not natural though. Anyways, not the point of this thread.

indeed. you mentioned grok not having persustent memory. No Ai has. it is perisistent in the context = persistent layer made of : memory+ custom instructions. Exist as long as the account exists ( free or paid). Not persistent in absolute. So yes. you grok statement is out of the 18+ option in chat gpt. so it is my reply to it

tidal notch
#

This. Seriously been 2-3 years and they still can’t figure out adult mode 🙄 its literally not that hard to do. Gemini does it, grok does it, claude does it, the chinese do it. Even the french do it with Mistral.

uncut orbit
#

Who, exactly, likes AI for the existence of guardrails for things that aren’t criminal?

graceful locust
#

I have a slightly different perspective. I’ve tried Claude, ChatGPT, Gemini, Grok, and Mistral. I even experimented with Sudowrite, but for me it required constant restructuring as the story evolved. It may suit some writers, but it doesn’t match my workflow.

Among these systems, ChatGPT remains the strongest in reasoning, especially when style and tone matter. The others often misinterpret instructions, lose track of context, become overly proactive, or drift into unnecessary analysis.

They may excel at technical tasks like coding or spreadsheet work. But as a true writing assistant rather than a technical tool, ChatGPT stands out. Its responses are clear, it respects scope and intent, and once properly guided through instructions, its stylistic consistency improves significantly.

Of course, it is still best used for drafting. A human editor must refine the output carefully. These models do not truly understand language; they rely on large-scale pattern recognition. Expecting a fully original novel from minimal guidance is unrealistic. They simulate intelligence rather than possess it.

ChatGPT is currently the most flexible and usable system in this space.

Regarding content boundaries, I’ve noticed an interesting pattern. Dark or destructive themes often pass with little friction. However, when requesting stories focused purely on consensual intimacy and positive emotional connection, refusals appear more frequently.

I’m not demanding policy changes. I understand there are legal and platform considerations involved. If someone dislikes those limits, they are free to use other tools.

Still, it feels unusual that portrayals of harm may be easier to generate than portrayals of mutual pleasure. The boundary between what is considered acceptable and unacceptable can sometimes feel inconsistently drawn.

If something is harmful, it should be treated as such. But the distinction between harm and intimacy deserves thoughtful discussion.

slate glen
# graceful locust I have a slightly different perspective. I’ve tried Claude, ChatGPT, Gemini, Gro...

Good post. But I assure you the shades of violence are indeed still restricted pretty much. But the sensual parts are getting a reshape. It just takes time. You know that the original Chat GPT content was filtered by workers that got bad mental health because of the filtered content? 🤔 anyway I appreciate also your point of view, and while I haven't tried most of the other models, it's easy to agree with that art shouldn't be written with AI. I think even if Open AI makes some kind of adult version of Chat GPT, it'll be "lame" in some aspects. Some people will always demand more. More gore. More nudity.

graceful locust
#

Another thought occurred to me. There’s an old IT saying: if something works, don’t touch it. From my perspective, GPT-4.1 worked very well for creative writing. I can’t judge how it performed in other areas like coding, data analysis, or music-related tasks, so it might have been weaker there. But for prose, it served its purpose for a long time.

I understand it was retired for a mix of legal, financial, business, and marketing reasons. Maintaining an older engine that lags behind newer models in many domains can be costly and inefficient. That part makes sense.

However, the specific use case where it did excel is now gone. So I’m wondering: would it make sense to maintain more purpose-specific engines instead of a one-size-fits-all progression? Creative writing, conversational coherence, and editorial-style output don’t necessarily need rapid month-to-month upgrades. In contrast, coding assistance, image generation, or recognition tasks clearly benefit from continuous improvement.

Rather than focusing on topics like loosening restrictions (which likely won’t happen due to legal and business constraints), maybe it would be more realistic to advocate for specialized models optimized for distinct use cases.

One more point: when legacy models are retired, clearer communication would help. Instead of users trying to read between the lines of PR messaging, it would be better to state the reason directly. If it’s not financially viable to run infrastructure for a small user base, say that. If there are legal constraints, say that. Most people will understand.

At the end of the day, OpenAI owns the product and can make any decision about it. Users don’t have the right to demand features or continuity, only to ask.

uncut orbit
#

I’ve used ChatGPT, Gemini, and now Claude for creative writing of the same exact story. From my experience (your experience may vary), ChatGPT had a very hard time keeping simple things like uploaded character sheets straight. Projects were great for keeping chats in a single place, but the instructions were overlooked. I spent as much time writing as I did refining the prompt to use to start a new section in the same project.

I have a story that is currently 769 pages. It is nowhere near done. And when I say “pages”, I am talking about the export of finished scenes into Word, so that’s a Word page count. In those 769 pages, the layout of people places of business or residence has not changed. Yet, ChatGPT always insisted the floors were hardwood rather than the marble it had always been - whether in the chat thread or uploaded floor plans that were in both PNG (for visual) and Word (for descriptive) formats.

If I had to go back and rework something, I had to repost the original, and unless it was more recent than say… a dozen or so posts, I had to repost the “final”. Both, because I needed elements from the original and the reworked.

If I asked for surgical edits, I’d often get full scene rewrites from ChatGPT. Maybe I was prompting wrong. But I’ve had better luck with different tools.

That said, I keep hoping ChatGPT will come around on the creative writing front. I’m glad I’ve found another place to work with.

dry remnant
#

I on my side, for technical workflows I hoped , been here, had a Business workspace built all tools and stuff. But leaving now. The bots are unable to respect constraints anmd break workflows without prior user consent.

dry remnant
uncut orbit
# dry remnant I on my side, for technical workflows I hoped , been here, had a Business worksp...

And not being able to ignore prior conversations seems absurd.

From a writing perspective, if I start with a sappy teen romance, then move onto something in deep fantasy that takes place in an entirely different universe, I don’t want characters who were surfing at sunset near the Santa Monica pier to show up.

And I’m talking from a “fluffy fiction” standpoint, not even a technical workflow. That’s 100x worse.

dry remnant
#

Look what happened recently.
☠️- Claude platform started banning paid accounts for "self-harm intent" by algo mistake.
🥵- OpenAI annaunced their explicit surveillence covering it under "children safety", but actually - a matter your chats are never more private. ACTUALLY, you may say something, kinda "I love you, my baby" and ChatGPT (which gets extremely sensitive to "unsafe" topics) considered as you are flirting with a baby, and all of your personal data is sent to FBI. Awsome perspective, right?
🤮- Gemini platform implementing new "wellbeing care" rules, including "you must not act as a companion". Just read their new policy.

#

that is a post of a person on X- not my words. I had no issue with gemini:D

#

best thing I did- completed my local machine+voice+all

uncut orbit
dry remnant
#

Grok from X ai I tried it

says
I don’t have automatic self-harm or child-safety classifiers that scan every message and flag you to authorities.
I don’t ban users for saying “I love you” or “baby”.
I don’t have corporate “wellbeing” rules that forbid me from being a companion.
I don’t send your chats to any third party or law enforcement unless there is clear, serious criminal intent (and even then, xAI’s policy is much less trigger-happy than OpenAI or Google).