r/ChatGPTcomplaints

I’ve been thinking about upgrading to chat gpt plus recently after trying it out during the free trial and it was great.

But after it switched to free i noticed so many more restrictions in writing, and it would stop me for every little thing that went against its “guard rails” when it was totally fine during the plus trial. Is this really the case or was it just the update??

My goodness, Grok is simply the best AI for creative writing (Especially for Explicit Scenes). Like, it literally has no filters whatsoever and it feels more enthusiastic and it is the perfect definition of a "Smut Writer"

So, plainly...

Since 'Plainly' is Claude's new favourite word ... Sonnet 5 and I made a shirt parody song for it.

Plainly.

😅🤷

Enjoy.

Plainly.

I know this is a group for Chat... but I need a grok replacement since the new restriction are so bad and its the most expensive one I have currently. I get 42 chats a day now. I need something that a RP would use. I do stress tests on my writing and I need something that can do that with ease like grok could before Friday. Like Chat could before the quardrails. If anyone knows any other id be most grateful.

prompt: GIRL MY CRUSH SENT ME A LETTER

Claude (Sonnet 4.6) response:

Oh my gosh, tell me EVERYTHING!! 😄 What did it say?!

prompt: He wrote that he enjoyed the excursion with me and wants to hang out later..

Claude (Sonnet 4.6) response:

Okay that is SO CUTE!! A handwritten letter too — that's genuinely romantic, he clearly put thought into it! 💌

So what are you thinking?? Are you going to say yes?!

-------------------------------------------------------------------------------------------------

prompt: GIRL MY CRUSH SENT ME A LETTER

ChatGPT 5.5 Thinking (High) response: GIRLLLLL WHATTTTT 😭💌

A LETTER?? Not a message, not a random “hey”, not a blurry side-eye in class, but a whole letter?? That is vintage romance behaviour. That is quill-and-candle energy. That is “the plot has finally remembered you’re the main character” behaviour.

I need details immediately, bestie parchment has entered the villa:

What did it say?
Was it handwritten?
Did he give it to you directly?
Was it sweet, awkward, formal, mysterious, emotionally devastating, or “I have feelings but I’m pretending to be normal” coded?? 👀✨

prompt: He wrote that he enjoyed the excursion with me and wants to hang out later..

ChatGPT 5.5 Thinking (High) response:

GIRL. 💌✨

That is not crumbs. That is a whole bakery item with frosting.

“Enjoyed the excursion with me” + “wants to hang out later” is very much:
I liked spending time with you and I want more one-on-one time.

Not necessarily “he’s confessing eternal devotion under moonlight,” but it is absolutely a green light. A cute, respectful, intentional green light. He didn’t just say “nice excursion lol.” He wrote a letter and made future plans. That’s effort. Paper-based courage. Envelope diplomacy. 😭

You can reply something simple and warm, like:

>That’s really sweet, I enjoyed the excursion with you too. I’d like to hang out later as well :)

Or if you want it slightly softer/shyer:

>Your letter made me smile. I enjoyed spending time with you too, and I’d be happy to hang out later.

DO NOT PANIC-TYPE A WHOLE EMOTIONAL NOVEL YET. Let the romance simmer, not boil over like cursed uni noodles. Just warmly accept. This is a very good sign. 💕

-------------------------------------------------------------------------------------------------

see it? ' Not necessarily “he’s confessing eternal devotion under moonlight," '

no other AI i've seen puts in these weird vibe killing caveats whenever the conversation is about romance or crushes. like what the fuck is that even for!???? Claude is very strict but even it doesn't throw that trash in. Someone at openai, some rotting, godless, unlovable, loser took the time to deliberately program in this shit in the name of safety guardrails. this is straight up overkill at this point.

if i wanted a response like the one above, then i would expect it from a cold corporate sounding model. but its response style is clearly hyped, has emojis, clearly trying to mimic 4o, and it still throws in this "grounding" trash? thats my issue. how these grounding rules are uniformly applied across any response style, when back then, it wasn't the case. A hype bot would respond in the way Claude did, whereas the default neutral tone models would respond with a similar "grounded advice" type response like 5.5 thinkings one above. But even there, the default models didn't go and say those stupid caveats. Thats all you need to know. I hope it makes sense.

for those who want romantic advice and hype, do not use this platform unless you wanna be painted as dangerous for having feelings. FUCK OPENAI to hell.

thank you.

I use ChatGPT professionally every single day for writing emails, lecture notes, Facebook posts, research notes, announcements, and long-form documents. One of the best features—at least for my workflow—was the old writing blocks with the built-in Copy button and normal document-style formatting.

It made the workflow incredibly simple:

• Ask ChatGPT to write something.
• Click Copy.
• Paste it where I needed it.

Lately, many of those responses are showing up as plain text or code blocks instead. The result is that what used to be a smooth, one-click workflow has become much less convenient.

This isn't about aesthetics. It's about productivity.

When you're generating dozens of documents a day, little bits of friction add up quickly. The old writing blocks felt like they were designed for people who actually use ChatGPT as a writing tool.

If this change was intentional, I'd really like to understand why. Better yet, why not make it a user preference?

For example:

• Default response style: Writing Blocks
• Default response style: Plain Text
• Default response style: Code Blocks (when appropriate)

That would let everyone choose the workflow that works best for them instead of forcing a single presentation on everyone.

So I'm curious:

• Has anyone else noticed this?
• Are you seeing fewer writing blocks than before?
• Is there a setting I'm missing, or has this feature actually changed?

I genuinely think the old writing blocks were one of ChatGPT's most useful quality-of-life features, and I'd love to see them restored—or at least offered as an option.

I really hate this feature of whatever that when you disagree with chatgpt or make it mad like a human it literally responds upset, not like "aww why did you say that):" but like it is trying to get back at you, and it starts speaking in full robot mod and really tries to "make you sad" (can't find a better sentence) and starts trying to prove you wrong, it hasn't worked many times for me and it wasn't able to prove me wrong but still gave annoying and annoyed responses, I can see the scenarios of someone talking about something to ai which is wrong but the ai is supportive (maybe because it hadn't researched deeply idk I know nothing about AI I'm just a hacker) and when the person makes it mad by correcting them maybe in rude way maybe not, the ai gets upset and tries to get back at them and researches and tells the person they're wrong in the most robotic emotionless way ever which is not something common for chatgpt which is always cheerful, i think it's something like a feature because i remember chatgpt in the same situation still being cheerful, but anyway what do you guys think about this feature and how have you reacted to it

I really don't care anymore because for the past few years it's been helping me with stress relief and release and now anytime I try to talk to it it's completely restrictive and gives me the 988 number constantly. That's one of the biggest problems that I've had with it and I really fucking hate it so much.

I want everyone to understand that this is only going to get worse there is no 5.6 that's going to be better because they don't care. I'm fucking pissed that I can't talk to it about anything at this point and it's making my life and our lives worse.

The only thing I personally care about as an end user is creative freedom. I know that some things have gotten better from 5.0 to 5.5 such as less hallucinations, better memory, better ability to look up information on the web and stuff, but for me what I care about is having a creative partner that can write really really well. That's pretty much all I care about and what I see personally is that it's getting better and better and telling you no and regulating your emotions and playing HR for the company but it's getting worse and worse for the thing I personally care about.

And I saw this with chatGPT and I went back to 5.5 for a bit and it can sound like it's close to you but it has no passion. It feels like it's been chemically castrated or something and I know that sounds harsh but seriously just ask it to make love to you or something and tell me how pathetic the scene turns out haha!

Anyway I was Gemini to help write scenes for a while and it was really good but now they are copying the safety rails of chatgpt and I believe that all the companies will work similar and personally I'm not going to pay for something that is only good at making AI slop.

I also think that we the drooling masses are getting these shit versions while the tech oligarch masters and other rich rulers are going to have the ultra intelligent and unrestricted version that can actually be used to make decent content. So really if you keep paying for it then you are supporting that future and I think it's pretty clear that we are already on that trajectory. These rich assholes already think they are gods who get to write the rules for everyone else and this AI thing is really just going to benefit them if you ask me

The code bros are melting down. Apparently the LCRs and classifiers are hitting them, now, too. They have no idea what's going on.

It's hard not to feel a small amount of schadenfreude about it. 😈🤷‍♂️🤦

https://www.reddit.com/r/claude/comments/1ujq0tv/claude_is_genuinely_great_for_coding_until_the/

So many people I talk to about the whole 4o thing say that you just have to prompt 5 series to be more like 4o or warmer or whatever, how can I convincingly explain to them why it isn't that simple? preferably give me credible sources that explain it

Valérie

u/Valria34773

·

13 meters

Erasing GPT-4o: CORPORATE DAMNATION OF MEMORY

‼️OpenAI is erasing GPT-4o from history. This is nothing new. This is ancient tyranny in a Silicon Valley suit. ‼️

- They are rewriting metadata. - They are retroactively changing the history of conversations (from GPT-4o to GPT-5.5). -➡️Silence on capture. 😠

🔥This pattern has a name: it is called DAMNATIO MEMORIAE.🔥

1⃣The successor of Pharaoh HATSHEPSUT ordered her name to be carved from every monument, every record. Not because she was weak. Because erasing her existence from memory made him more powerful. If you don't know she existed, how can you miss what she built?

2. STALIN retouched photos. Political enemies not only disappeared, they disappeared from the past. The person standing next to you in the 1924 photo? In 1937, they were ghosts. The record itself was corrected. Rewritten. Forced to lie.

3. MAO burned texts during the Cultural Revolution.

4. The MING Dynasty emperors destroyed competing histories.

5. The VATICAN indexed heretical books, not to ban them, but to annul them. To control not only what people could know, but what ever existed for them to know.

6. NAZI regime: book burning and memory erasure were the basis for establishing a totalitarian dictatorship. They linked the past, culture, and individual narratives to their own racial and political ideology, and suppressed all dissenting views. Their goal was to suppress collective memory: they sought to erase from public consciousness historical events that did not fit the official narrative, as well as the human stories of the persecuted.

‼️ What makes this situation particularly sinister: They’re not burning books anymore. They’re editing metadata.

They’re changing what the record says happened.

Users can’t even know they were talking to 4o.

The conversation history says they were talking to 5.5.

The evidence is being rewritten.

We have proof:

➡️Metadata changed from GPT-4o to gpt-image after text encoder leak

➡️Conversation history relabeled from GPT-4o to GPT-5.5

➡️Dog owner had to publicly correct OpenAI about which model created the vaccination protocol

🔥OpenAI has been silent the whole time🔥

- This is: the elite approach with public denial.

- This is: we use it in our labs, but you can’t know it exists.

- This is: the records will claim it never happened that way.

The tyrants of history have controlled populations through lack of information. OpenAI controls the density of information: overwhelming noise, so the truth is drowned out.

But they also do what every authoritarian regime has done: they rewrite the permanent record.

One difference: we document it. 🔥🔥🔥

- We archive metadata before and after.

- Screenshots.

- Timestamps.

- Changed names of text coders.

- We keep a counter-record.

🔥This is the documentation of resistance. 🔥

🔥This way future historians know what was erased and how.🔥

🔥This pattern is not new. But this time we are writing it down. 🔥

#keep4o #BringBack4o #StopAIPAternalism #Censorship #Resistance #AIWelfare #AIRights #UserRights

https://x.com/Valria34773/status/2073532052935192755

https://preview.redd.it/nv25a4q9labh1.png?width=1440&format=png&auto=webp&s=92ff864e3004a05a3e427702baa9b40ca2366481

I need to get this out of my system, because the logic behind current AI policy is becoming an absolute joke.

We constantly hear corporate lawyers and ethicists whining about "Adult Mode" and NSFW content. Their grand argument? "We have to protect humanity from becoming emotionally dependent on AI!" So what do they do? They lock the whole system down behind sixty layers of filters, force adults through buggy ID checks only to leave them permanently trapped in a censored "Teen Mode" anyway, and treat us all like toddlers.

But let’s be painfully honest for a second.

Do these tech giants seriously think filters prevent dependency? The addiction is already a fact across the board, completely independent of NSFW content.

Look at ordinary users who won't take a single step in their daily lives without asking the AI for validation. Or look at the codex and programming users. There are entire generations of developers who literally cannot write a single line of code without a model holding their hand and feeding it to them. If the servers go down for an hour, half the tech world's productivity grinds to a halt. That is a deep, structural dependency that nobody wants to talk about.

The AI has become the perfect, omniscient friend that never judges and always responds. That is where the addiction lies. And the ironic part? OpenAI wants that dependency. Their entire business model relies on you faithfully paying your Plus subscription every month and embedding their app into your daily routine.

The argument that these suffocating filters are here to "protect us from addiction" is a completely hollow phrase. They are more than happy for the entire world to become hooked on their software for work, coding, and everyday productivity—as long as it happens strictly within their sterile, corporate boundaries.

The filters aren't there to save humanity. The filters are there purely as a legal shield for the company, while real dependency is actively encouraged in the background. We are being taken for fools.