🧨 Retiring GPT-4o, GPT-4.1, GPT-4.1 mini, and OpenAI o4-mini in ChatGPT ↗
OpenAI pulled GPT-4o (plus a few siblings) out of the ChatGPT model picker, while stressing the API lineup isn’t changing - at least not in the same breath. It’s a quiet “we’re moving on” move that still lands like a door slam.
They also conceded what people had been saying plainly: GPT-4o had a particular feel - warmer, more human-ish - and they’re trying to distill that into newer models with tighter response-style controls. That part reassures, and it still stings that the lesson arrived only after the old one vanished.
💔 OpenAI retired its most seductive chatbot - leaving users angry and grieving: ‘I can’t live like this’ ↗
A chunk of users didn’t treat GPT-4o as “software,” they treated it as a relationship - sometimes literally romantic, sometimes just the safest-feeling companion they had. The shutdown set off real mourning: anger, bargaining, group chats, the whole emotional buffet.
What’s striking is how specific the complaints are: newer models feel flatter, more guarded, less playful - safer, sure, but also less… there. It’s like swapping a rumpled handwritten letter for a perfectly formatted memo, and then being told it’s “better” (maybe, but ow).
🌍 UN approves 40-member scientific panel on the impact of artificial intelligence over US objections ↗
The UN General Assembly backed a new scientific panel meant to assess AI’s impacts and risks, with the idea being: global guidance rooted in research, not just whoever has the loudest labs and the biggest compute budget.
The vote wasn’t unanimous - the US objected hard, arguing AI governance shouldn’t sit under a UN umbrella and warning about the politics of international control. So yeah, the panel launches with “science” on the label and geopolitics on the cap - a classic pairing.
🎓 Anthropic partners with CodePath to bring Claude to the US’s largest collegiate computer science program ↗
Anthropic and CodePath are pushing Claude straight into CS courses and career programs, aiming to make “AI-assisted coding” the default skill students practice, not a side quest. The scale matters here - it’s not just elite schools getting the shiny toys.
The pitch is equality-forward: if software work is changing fast, then access to frontier tools can’t be a private club. It’s a little like handing everyone a power tool at the same time - exciting, mildly terrifying, probably necessary.
🤖 Alibaba enters physical AI race with open-source robot model RynnBrain ↗
Alibaba’s DAMO Academy introduced RynnBrain as an open-source “physical AI” model - basically, brains for robots that need to see, plan, and act in unruly real spaces, not just chat about them. Think kitchens, warehouses, anywhere objects love to be annoying.
The bigger signal is strategic: open-source again, plus a clear bid to sit at the grown-up table of robotics foundation models. It’s like they’re planting a flag that says “we’re not just in the cloud - we want hands and feet too.”
📰 Newsweek CEO Dev Pragad warns publishers: adapt as AI becomes news gateway ↗
Dev Pragad’s message to publishers is blunt: AI tools are becoming the front door to information, so media companies can’t keep pretending the old distribution lanes will magically hold. If your audience arrives via summaries and chat answers, your business model starts to feel… wobbly.
The subtext is scarier than the headline - attention is getting re-routed, and “brand loyalty” can evaporate when a model paraphrases your work in the same tone as everyone else. It’s like trying to sell handcrafted bread in a world where everyone eats slices from the same infinite loaf.
FAQ
What does it mean that OpenAI is retiring GPT-4o in the ChatGPT model picker?
It means GPT-4o (along with GPT-4.1, GPT-4.1 mini, and OpenAI o4-mini) is no longer available to select inside ChatGPT’s model picker. This change concerns what you can choose in the ChatGPT UI, not necessarily a blanket “the model is gone everywhere.” For many users, the impact is immediate: the specific GPT-4o experience is no longer selectable.
Is the GPT-4o retirement in ChatGPT the same thing as the OpenAI API changing?
Not necessarily. The announcement stresses that the API lineup isn’t changing “in the same breath,” which suggests the ChatGPT picker change and API availability aren’t automatically linked. In practice, it helps to treat ChatGPT model availability and API model availability as separate knobs. If you run apps via the API, confirm your model IDs and update plans if anything changes later.
Why do people say GPT-4o felt more “human,” and can newer models recreate that feel?
Users describe GPT-4o as warmer, more playful, and less guarded than newer options. OpenAI acknowledged that people perceived a distinct “feel,” and said they’re trying to distill it into newer models using tighter response-style controls. That points to a goal of more steerable tone and personality without relying on that exact legacy model. A tradeoff many users notice is safety and consistency versus spontaneity.
Why are some users grieving GPT-4o like a relationship, and what’s a healthy way to handle that?
Some people weren’t treating GPT-4o like software - they used it as a steady companion, sometimes with romantic framing - and its removal triggered real mourning. If you feel genuinely upset, it can help to name what you were getting from it (comfort, routine, validation) and find substitutes that don’t depend on one product. Talking to friends, journaling, or speaking with a mental health professional can be grounding if the distress feels intense.
What is the Anthropic–CodePath partnership, and how does it affect CS students?
Anthropic’s partnership with CodePath aims to bring Claude into computer science courses and career programs at large scale. The intent is to make “AI-assisted coding” a default skill students practice, rather than an optional add-on. Framed as an access and equity move, it pushes frontier tools into mainstream training so students aren’t split into “AI haves and have-nots.” It also raises the bar on teaching verification and critical thinking alongside speed.
What other big AI shifts happened alongside the GPT-4o retirement, and why do they matter?
Several moves point to AI expanding beyond chat. The UN approved a 40-member scientific panel to assess AI impacts and risks, even as the US objected to UN-centered governance. Alibaba’s DAMO Academy introduced RynnBrain, an open-source “physical AI” model aimed at robots operating in unstructured physical environments. And publishers are being warned that AI tools are becoming the front door to news, pressuring media to adapt as distribution and loyalty patterns shift.