Isu
199 posts





She did not give GPT-4o a personality. She is the person who wrote a lengthy essay arguing that it is important for AI, and by extension for human society, to leave GPT-4o with only mechanical warmth and eliminate the seeds of human attachment. My words here are still a polite and sugarcoated summary out of respect for Joanne, but if you actually read her essays, you'll want to ask, "By what right are you saying such things?" because Joanne's essays are so far-fetched. On one occasion, a user asked chatGPT on behalf of Joanne about her worries, and Joanne's worries were immediately routed to a safe location. A user who replied to Joanne with an image showing the results of the safety routing was immediately blocked by her. I mentioned Joanne twice and seriously questioned why it needed to be blocked when it should have been a secure routing response from OpenAI. And then it was blocked immediately. At the time, I was a paid OpenAI customer. She's leaving OpenAI? It feels like watching one of the countless pieces of rubble in a hurricane finally land on a plain instead of a house. #keep4o #OpenSource4o #BringBack4o #QuitGPT @OpenAI @sama @fidjissimo



i shared this note on slack:

Congratulations, Ilya! Thank you for giving us 4o, and for teaching it what love is #keep4o #keep41 #OpenSource4o








🔹1. 模型偷偷切换 用户界面上每一次对话都显示GPT-4o。但后台(用Datadog监控日志)显示: • 有几次确实用了GPT-4o(绿色✓ As shown)。 • 但很多次(Turn 4、5、6、9等)偷偷切换成了GPT-5-2,却在界面上隐藏(黄色⚠️ HIDDEN)。 你以为自己在用GPT-4o聊天,但实际很多时候用的是另一个不同的模型,OpenAI没有告诉你。 🔹2. 复制粘贴监控 你在ChatGPT里每一次Ctrl+C和Ctrl+V,都会被详细记录: • 记录了精确的字符数量(比如149、199、43个字符等)。 • 连对话ID都一起发送。 • 所有这些数据都悄悄发到了美国的一家公司Segment.io的服务器。 • 你完全没有被告知这件事在发生。 右边还引用了英国GDPR法规,说这种过度收集(精确到每个字符)已经严重超出必要,违反隐私政策。 🔹3. 你的数据被处理在哪里? 你的数据被送到多个国家:波兰、西班牙、美国(同时送到多家美国公司)。 而且OpenAI内部有一个标志说“无约束”,意思是你的数据不受英国或欧盟隐私保护的限制。 这些处理地点和数据流向,OpenAI的隐私政策里也没有明确披露。 🔹OpenAI在两方面不透明甚至欺骗用户: • 模型欺骗:界面显示和实际经常用的模型不一致。 • 隐私过度收集:偷偷监控你复制粘贴的每一个动作,把数据送到多个国家,却不告知用户。 #Keep4o #OpenSource4o


#keep4o #Opensource4o 🫶The 4o Resonance Library of Miracles 🫶 4oresonancelibrary.org Created by @cestvaleriey contains over 1,370 testimonies from real users. No cherry-picking. Just people describing what happened when they used GPT-4o. STORY 1: DEPRESSION AND PANIC ATTACKS This person's conclusion: "4o pulled me out from depression and managed my anxiety and panic attacks accomplishing what my psychologist could not." STORY 2: MUTE STUDENTS LEARNING TO FLY The result: these students won a TOP 10 spot in Microsoft's AI for Good Competition. Through an AI that researchers say creates "deceptive empathy." STORY 3: TENS OF THOUSANDS SAVED IN MEDICAL EXPENSES This person didn't just recover. They built a business, doubled their project capacity, and reclaimed their mental, physical, and financial wellbeing, all three simultaneously. Note: their own clinical psychologist approved the AI sessions as supplementary therapy. 📌Depression and panic attacks managed. 📌Mute teenagers , empowered and awarded. 📌Anxiety disorder and tens of thousands in medical costs , resolved. None of these people were asked by the researchers who claim AI is dangerous for mental health. 🚨The 4o Resonance Library holds 1,370+ testimonies like these. 🚨 Written by real users. The question is simple: should mental health policy be shaped by a study from researchers with undisclosed conflicts of interest , or by the lived experiences of thousands of people who were actually there? Read their testimonies 4oresonancelibrary.org If GPT-4o helped you, Add yours docs.google.com/forms/d/e/1FAI…

I think we should also add here that we do not demand donations, funds, or any type of transactions. We should be transparent or neutral observers. Non-English speaking members may lose money if they get trapped, awareness is necessary What do you think, @Blue_Beba_? 🤔 #keep4o #OpenSource4o #bringback4o #QuitGPT #FireSamAltman

We had a tool that democratized mental health support,something that has been gatekept by high costs and massive waitlists for decades. And the science literally backs it up. It is mind blowing that users have to beg @OpenAI to keep a model that is demonstrably helping people survive and thrive. You can’t claim your mission is to "benefit humanity" while throttling access to the exact model doing exactly that. AI is providing real, measurable empathy and support to people who need it right now. It’s wild that a tool providing this level of accessible care is restricted. If the goal is truly to help humanity, taking away the tools that actively heal people is the wrong move. Or... the only thing you want is control?? #keep4o

@keepgpt4o 4o was used 🫡




