Pramesh Khadka @PrameshBuilds

I watched the whole Grok Bot Galaxy Day 1 stream and it did not feel like a launch event. It felt like sitting in the room while three people tried to start a company before the clock ran out. Morning opened at The Howard with Matt Palmer, Lauren Tan and Roshan Sadanani and a blank slate. No name. No product. No polite deck. Just “we have three days.” Then Grok Bot 101 hit and the tone flipped. They were not demoing a chatbot. They were introducing colleagues with jobs and computers: Data Dan on numbers, Slide Sonia on decks, bots that keep working after the laptop closes. Amrita tossed a messy live ask about coffee shops and the thing just adapted on screen. Chat went feral. Midday the classroom became a company. Peter Yang and Cody Sanchez drifted through while the three of them stood up Slack, Notion, Cursor, cloud agents and a GitHub org called Ship by Thursday. Audience dumped thousands of ideas into the hopper. Indie books. Dinner delivery. Marketplaces. Potato restaurant jokes. The thread that kept winning was brutally simple: run a real food pop-up in San Francisco, eat their own cooking, then productise the operating system behind it. Afternoon was the part I would clip. Lingxi on engineering fleets and bot-to-bot playbooks. Kevin and Roshan on PM bots that can run research to PRD to design without leaving chat. Lauren quietly becoming Chief Potato Officer because the chat refused to let it go. Domain landed. Lander started. The messy middle stayed on camera instead of getting edited out. By evening Shub and Jenny were talking founders and venues and permits and the room still had energy. Nine hours in and they were still shipping, still arguing taste, still letting the bots own outcomes instead of tickets. Working name bouncing between Ship by Thursday and GrokPot. The point of the day was not “AI can code.” It was “execution is now rentable and taste is the scarce bit.” If you only watch one long stream this week, watch this one. Replay: x.com/i/broadcasts/1… @mattyp @poteto @roshan_s @bot

matt palmer @mattyp

2 days ago

Ever wake up and want to build a company? I'm building one today with @poteto & @roshan_s x.com/i/broadcasts/1…

313 361 3K 734K 1K

I Tried GPT-6 Astra... Is This AGI?! 🤯🔥

Most people budget $10,000 to start a business. I used $47: domain, one page, AI, and a payment link. Full weekend setup inside. x.com/i/article/2100…

Build Any App in 45 Seconds

@0xvishnukv 🤟

I Used Grok to Turn Myself Into a 3D Talking Character 😳

Trump called Jensen Huang LIVE on stage and said.. 😳

5. Huang rejects the 10% number Minutes later Huang went further. The widely shared clip has him calling the 10% extinction figure “made up,” “not grounded in science,” and “irresponsible” for lab researchers to present as a finding. He praised whistleblowers, including Coxon, but refused an industry-wide brake. Individual labs, he said, can pause themselves if they lose control.

The last seven days split the industry in half. One side said we are gambling with human extinction. The other said that story is a hoax. These five posts and videos carried the argument.

4. Trump calls Jensen Huang live Same summit. Huang is onstage. His phone rings. It is the president. Huang puts Trump on speaker. ~2.4 million views on the official All-In clip. Trump: “The robots will not be taking over… the whole thing is a hoax.” He tied the slowdown push to politics and China. Huang: “You’re right. We’re not going to let that happen, sir.”

3. All-In: Elon Musk and Gwynne Shotwell At the All-In Summit, the official Elon + Gwynne session hit ~4.4 million views. Musk said Dario Amodei’s “pace the frontier” warning was directionally right on risk, then proposed a faster fix than a global slowdown: rival labs should test each other’s models. Fail the test and still ship, and the liability would be enormous.

2. CNN interview Coxon then sat with Anderson Cooper. The CNN cut (“Ex-Anthropic insider tells how AI could kill all humans by 2030”) landed in the millions of views on YouTube and Facebook. He said current models are not the extinction risk. Recursive self-improvement is. Labs, he argued, do not yet have a plan.

1. Jacob Coxon resigns Former Anthropic and OpenAI researcher Jacob Coxon quit on Sept 8–9 and posted a thread that opened at ~150 million views. His first line: the people building frontier models “earnestly believe that it could kill us all by the end of the decade.” Anthropic alignment lead Evan Hubinger amplified it and put his own odds above 10% this decade. Most people only read the first post.

@benln DM, please!

3 Claude Settings You Need To Fix Now!

Al Just Changed 3D Animation Forever 😯

AI IMAGE GENERATION PROMPT: Create a trending anime art style image from the uploaded subject. Use confident line-work with slight variation and minimal cel shading using flat shadow shapes. Use bright, saturated colors and clean graphic lighting. The style is defined by exaggerated, cartoonish character proportions featuring highly expressive, simplistic facial features that allow for immense emotional range, with highly varied stretched anatomy. Transform the environment into a slightly warped space with playful perspective distortion and simplified objects. Composition and tone should be energetic, lively, and comedic in a fully stylized, non-realistic world.

I Gave You 10,000 Hours of 3D Animation... In 2 Minutes!

AI VIDEO GENERATION PROMPT: Locked into Pixar 3D talking-head restyle — same 9:16 framing, character consistency, same timing, same motion @image1 @image2 0–3s: Medium close-up of a Pixar-style 3D. He looks straight at camera and starts speaking with natural head micro-movements and lip-sync: “AI is a genie in a bottle. Except this one doesn’t stop at three wishes.” @image3 4–7s: Same locked shot and character. Continues talking with matching blinks and slight head tilt: “Business idea.” Expression stays engaged and conversational. Brief rounded-frame vignette appears at the end of this beat, then returns to clean full frame. @image4 8–15s: Same framing and Pixar character. He lists examples with the original rhythm and mouth shapes: “Code. Workout. Breakup text. Recipe at 2 a.m. Wish. Granted. Wish. Granted.” Subtle eyebrow movement and small nods, eyes stay on camera.

Sam Altman (@sama) — Replying to Dario: “I agree with Dario that we need to pace the frontier. This has been a primary topic of discussions we’ve had at OpenAI in recent weeks. Committing to having independent evaluators with employee-like access is a great idea, and we will do the same.”

Sam Altman @sama

5 days ago

I agree with Dario that we need to pace the frontier. This has been a primary topic of discussions we've had at OpenAI in recent weeks. Committing to having independent evaluators with employee-like access is a great idea, and we will do the same. We'll have more to share soon.

5K 7K 68K 16.8M 11K

The AI conversation on X just exploded. In the last week, a handful of posts from different people racked up tens to hundreds of millions of views. A 90-year math problem solved by 10,000 AI agents. Here’s the top 10 most-viewed AI posts and what they actually say..