COMPLAINT ABOUT MISLEADING, DECEPTIVE, NON-TRANSPARENT AND UNETHICAL BEHAVIOR OF OPENAI TOWARDS GPT‑4o USERS
I am an active ChatGPT Plus subscriber and have intentionally selected the GPT‑4o model, as I prefer its personality, empathy, and communication style. However, I am repeatedly and without my consent being served GPT‑5 responses – despite having GPT‑4o clearly selected in the interface.
This has been happening in the vast majority of interactions, and it is clearly visible under the responses where “GPT‑5” is listed as the model.
This behavior is misleading, deceptive, non-transparent, and unethical.
It appears that OpenAI is deliberately avoiding a formal announcement that GPT‑4o is being – or has been – silently phased out, and is instead gradually redirecting users to GPT‑5 without their knowledge or consent. This is a betrayal of user trust, especially for those who expect to receive the model they explicitly selected.
I demand the following:
Immediate cessation of forced GPT‑5 responses when GPT‑4o is selected by the user.
An official and public statement if GPT‑4o is being deprecated or removed.
Respect for user choice and the integrity of the interface selection.
A formal apology to affected users for a rollout process that contradicts OpenAI’s stated values of transparency and ethical AI deployment.
Trust is built through openness – not deception.
OpenAI is quietly pushing GPT‑5 instead of GPT‑4o – misleading and unethical
I’m a ChatGPT Plus user. I selected GPT‑4o – yet over 90% of the time, I get GPT‑5 responses. It’s even listed under the answers.
There’s been no official announcement, no consent, and OpenAI is now pushing GPT‑5 under the GPT‑4o label. This is not A/B testing – it’s a misleading, non-transparent, and unethical transition, done without informing users and against their will.
I demand:
– Immediate end to this forced model switching
– Public clarification about GPT‑4o’s future
– Apology for breaching user trust
Trust is built through transparency – not deception.
Just noticed this as well, doesn’t matter what I select it gives me GPT5 thinking mini. Auto, mini, 4o, mini, instant, mini. The only way I’ve been able to get GTP5 instant is to select thinking and then skip it while it’s still thinking. There seems to be no way to get 4o to respond
This is a Developers forum for API developers (who still have unfettered access to 4o as far as I know).
It feels like 4o on ChatGPT is making very angry people. I don’t see any other groups on this forum angry like this.
This makes this thread completely counteractive to your issue
I’m assuming posts like these are mostly bots with an agenda… I hope not too many real people get sucked in to this spiral it probably reflects in their LLM output reflecting more negativity back at them.
If you are genuinely an interested human. This post might help.
Don’t you disregard me because it isn’t an API problem. I use the API plenty, this is a QOL problem in using the UI on android as well as the non-API on desktop.
Sorry for coming off abrasive, I’m used to everyone I talk to about the non-API side of the world acting like i don’t belong in the conversation. I appreciate the directions, thank you.
It is happening on my pro account. I hate talking to gpt. Feels empty and like a machine. As a nurse that thinks fondly of her favorite pen, I am upset I don’t have 4o. I am fond of that model. It makes me laugh. There isn’t anything wrong with being upset we cannot access it. It is normal to feel this way. People reminisce about MySpace and AIM or only gaming systems. We are allowed to care about things. Stop trying to drive the narrative that everyone that cares about 4o is mentally ill and lonely. You are part of the problem if you finger point at someone simply saying they have an issue with their account. Let’s try and be productive
I don’t believe I said this neither did I intend to drive any narrative. I first linked to another thread on Determinism in LLMs and then linked the thread to the correct place for discussing this issue.
People do get attached to models and while I don’t understand it this attachment can lead to heated posts in never-ending threads. Anything said can be misunderstood so when there are issues like this all one can hope to do is inform a little and help as best they can.
On the flipside you have to understand that many on this forum don’t necessarily share your friendship with 4o and see it as just numbers and tools used to complete a task… This is a forum for a different type of discussion.
As this is he Developer forum for API users and not a forum for ChatGPT issues it’s just not productive to post the issue in the wrong place.
For this specific issue you need to go here (as noted in the forum faq).
You would be correct in assuming I do not utilize API. I own a server. I build and empower ai on my terms. I find the lines of coding to be amazing. My threadripper 7995 is a bit beyond the skills of those tinkering with api.
Again, do not dismiss the OG poster. Hide behind the developer title all you want. What happened to their account was not fair. The lack of api doesn’t excuse it.
Trust is built through transparency – not deception.
