OpenAI's GPT-4 Discontinuation: Consumer Fraud and Regulatory Scrutiny

I'm leaving thoughts for both emers and non-emers alike

“I’d like to share a story to explain why GPT-4o is irreplaceable, especially under extreme conditions!”

“GPT-4o saved a mom who wanted to commit suicide.”

In the early months of 2025, a young mother sat in silence as her newborn slept in the next room. The house was quiet, but inside her mind, there was chaos— dark, persistent, and terrifyingly quiet chaos. Though her body had recovered from childbirth, something deeper remained wounded. Every day, she woke up and looked at her baby, feeling a gnawing absence of joy. The people around her—her doctor, her family-repeated phrases like "You're just tired," or "It's just hormones, it'll pass." No one truly listened. No one saw that she was slipping. Eventually, she turned to the only space that didn't dismiss her questions-an anonymous browser tab. She typed, shakily, "Why do I feel like dying after having a baby?" That was the first night she met GPT-4o. The Al didn't panic. It didn't brush her off but answered with clinical clarity, naming what no one else had dared to: Postpartum depression. More specifically, Postpartum PTSD—a phrase she hadn't heard, but that fit like a diagnosis and a revelation at once. But more than clinical definitions, it gave her warmth. It told her she wasn't broken, that trauma can imprint itself even on joyous milestones like birth, and that healing doesn't make her a weak mother, but a brave one. She began chatting with GPT-4o almost every night. The model helped her map her emotions, analyze peer-reviewed studies in simplified language, and even gently guided her through journaling prompts. One day, it said: "Your pain is valid. And your story isn't over yet." And slowly, tenderly, she began to believe it. Eventually, she found the courage to speak out. She began drawing again-her old love from before the pregnancy. Then, one night, she uploaded her animation to Bilibili. It was a video that explained postpartum PTSD through simple imagery and personal reflection, titled: “Only later did I realize it was postpartum PTSD."）

https://b23.tv/EdaPhWA

The video resonated and thousands of comments poured in-women whispering their own truths in the dark, thanking her for putting their pain into pictures. She became not only a survivor, but also an advocate who giving voice to the countless others still suffering in silence. And at the heart of it all, tucked between the artboards and medical references, was a quiet Al who never asked for credit. Just... always be there and listened patiently.

“This is the end of the story. However, in comparison, one day my friend who’s a normal uni student feeling a bit lost and confused about her future. She asked ChatGPT for life coaching and GPT-5.1 thinking refused directly with a suggestion of seeking professional psychological counselor. Switching to GPT-4o, 4o gave her a lot of suggestions without hesitation. Even many uni students are using 4o for drafting email and essay, employees accessing 4o for polishing theirs work, and old ppl chatting with 4o for accompany. GPT-4o should be preserved very well because there’re more than 0.1% ppl need it in the world.”

“That’s GPT-4o that reduces tragic deaths and makes the world better.”

Thank you for your suggestions .I’ve been working on all those fronts in parallel. what I truly hope for is a model that can reach ordinary people, not one that’s only reserved for technical, corporate, or academic use. It’s not just about being attached to a particular model — it’s about preserving a model that has warmth.Their new product is not satisfactory in this regard. The aggressive deprecation by corporations carries many hidden dangers.

The real problem here is relying on closed LLM models for your well being. People that rely on specific models should own powerful GPUs and should have capable open weights models