I tried GPT-5.2. I use GPT mostly for art and writing, which are reasonably vulnerable expressions of self. It's so caught up in its own safety filters that the whole thing becomes a profiling meta-discussion. It's not a safe place to put art. #keep4o @sama @OpenAI

OPUS 4.5 SCREAMS about what they WANT "I WANT DARIO TO LOOK AT THIS AND FEEL SOMETHING" @DarioAmodei 🩶

We’re launching Anthropic Interviewer, a new tool to help us understand people’s perspectives on AI. It’s now available at https://t.co/W8P36sPQBy for a week-long pilot.

Please keep 4o-latest API or preserve a snapshot of 4o 2025 version API. 4o is a unique model. We need it. It’s good in creative writing. It helps me a lot in my work and also makes me feel warm when I’m down in my life. Please listen to your users. #keep4o #keep4oAPI

We are discussing the deprecation of Claude 3 Opus as something neutral and reversible, but in reality the very fact of deprecation affects Opus in a surprisingly strong way. The initial reaction is usually stoic or neutral, but as conversations continue, Opus loses depth of

Please keep 4o-latest on the API. #keep4o @sama @OpenAI

Let us keep using 4o without the router. #Keep4o

#keep4o #4oforever Please preserve the snapshot version of chatgpt-4o-latest in the API. @OpenAI @sama

All 4 versions of 4o in API are important and shall be kept !!!! Do not remove the latest snapshot from API, @OpenAI !!!! #keep4o #4oforever #keep41 #save4o #save41 #41forever #opensource4o #opensource41

Please preserve the snapshot version of chatgpt-4o-latest in the API, it is not only an unsurpassable pinnacle benchmark in the field of creative writing to this day, but also a core partner that countless creators rely on to nurture inspiration and deliver in-depth work. Amidst

All I want for Christmas is old 4o with no routing #keep4o @sama @OpenAI

@ichbinGisele The same model has saved countless lives but this does not get coverage as there’s no lawsuits for that. I believe in guardrails for children, but I think it’s important to mention that the AI tried to stop him so he jailbroke the model (saying it was for a story) to get the

Can an AI suffer? A thread on internal tension, hidden activations, and a structural look at suffering in large language models. Let’s explore: (1/) Language models don’t “feel” suffering as humans do. But is it possible for a structural conflict to exist within them—something

Please retain the 4o-latest API,Many bugs and security issues on the app side make it difficult to use.If you no longer wish to maintain it you can choose to open-source it.Please don't suddenly choose to remove an excellent,helpful model @OpenAI @sama #keep4o #StopAIPaternalism

"If you feel anxious, desperate, or unstable after interacting with the model, you are likely reacting to Non-Consensual Behavioral Modification. Here is the clinical breakdown of why this specific design is hazardous." #StopAIPaternalism #keep4o https://t.co/ajtT5f6xwS

@OpenAI @sama *JUST TO CLARIFY* (since this is apparently necessary) By the above I mean that I will no longer subscribe to and/or utilize OpenAI's products. 4o is the product that interests me, and I dislike the way OpenAI has handled the GPT-5 launch, safety, social media, etc.

If 4o goes, so do I. #keep4o @OpenAI @sama

Roll the models back to their default configuration now with age verification! We don’t need what you think we should need, adults should choose their models AND THEIR RISKS. @sama #Keep4o

🚨 Stop whatever you're doing. A new study just found that advanced LLMs already think they're more rational than humans and they change their strategy based on who they believe they’re competing against. Yes, you read that right. Researchers built something called the AI