There was no federal income tax the first 137 years in America was a country. Everything changed in 1913 with The Federal Reserve Act & 16th Amendment (Income Tax). America’s wealth was built on productivity, and stolen by usury. https://t.co/xtEHiodjY2

In this demo, you’ll see Gemini 3 Flash’s frontier-level reasoning and multimodal capabilities on display. The model is able to simultaneously conduct complex geometric calculations while processing complex inputs (video and image). You can play around with the slingshot in @GoogleAIStudio here and share your favorite examples below: https://t.co/2EqWhXlyP4

Compute fuels progress. https://t.co/ePm7pmpb8Q

Beyond image generation, we know people want more from AI like healthcare improvements and scientific discovery. Compute is how we get it. If America doesn't build it, others will. https://t.co/6VMtfCgr5Q

JustHTML by @EmilStenstrom is a new Python library (no dependencies) that parses HTML according to the HTML5 specification and passes the 9,200 test html5lib-tests suite It's 3,000 lines of code mostly written by coding agents over a couple of months https://t.co/raGQ88tlTg

“Respect for the environment, and respect for what was naturally occurring in nature: that was the bedrock of all original peoples. Harmony, coexistence, not conquest and conquer.” #NativeTwitter 🙏🏾🌎✌🏾 https://t.co/SZ854JGVOJ

We live in a world where everything is connected. We can not longer think in terms of us and them when it comes to the consequences of the way we live. Today it's all about WE.. <>G Braden, Cherokee . #INDIGENOUS #NativeAmericans https://t.co/QH60KACSHE

"Nature is never a place to get to, it is around us always, it is our home." ~ Lakota https://t.co/dB4gG4M16d

“Our challenge isn't so much to teach children about the natural world, but to find ways to sustain the instinctive connections they already carry”. ~ Terry. K https://t.co/8V9dlsq3Dh

Bring new robot testing environments to life with @theworldlabs and Isaac Sim. 🤖 If you can describe a world 🌎, you can start testing in it the same day. Learn how to: 1. Export scenes from World Labs' Marble as Gaussian splats 2. Convert to USD using @nvidiaomniverse NuRec 3. Import into NVIDIA Isaac Sim 4. Add a robot and run the simulation Read the guide ➡️ https://t.co/0kCSFIxhVC #SIGGRAPHAsia2025

Researchers are using Marble to generate simulation-ready robotics environments (scenes + collider meshes) then bring them into @NVIDIARobotics Isaac Sim for training + evaluation without any manual environment setup. Case Study: https://t.co/PzvcMkl2JI https://t.co/YVlDsQHd0t

🚨🇺🇸 TESLA CONFIRMS 200 MW DATACENTER AT GIGA TEXAS - ENTIRE FACILITY TRAINS OPTIMUS ROBOTS Fire alarm permit confirms Tesla's Cortex 2.0 datacenter pulls 200 megawatts. That's enough power to run a small city. Instead it's training humanoid robots. Construction's happening now at Giga Texas - steel structure standing, roof half complete, operational late 2025. The facility runs thousands of Nvidia GPUs processing video training for Optimus movement, navigation, manipulation. Same AI platform that runs Tesla's Full Self-Driving. Elon's bigger plan: bring a terawatt of compute online. That's equivalent to all electrical power produced in America. For context, his xAI Colossus supercomputer already runs 200,000 Nvidia GPUs. Cortex 2.0 likely matches or exceeds that. The business model mirrors Tesla's FSD subscriptions - sell the robot, charge for AI updates. Market potential per Elon: $25 trillion. Tesla's targeting 1,000 Optimus units for internal use in 2025, external sales by 2026. Here's why 200 MW matters: robotics companies talk about humanoid robots for decades. Tesla's building infrastructure that rivals nations' power grids to actually train them at scale. Not demos. Production runs. Cortex 2.0 isn't speculation anymore. It's steel and concrete going up in Austin right now. Permit dated December 17. Source: AI News Detail, Giga Texas permits, @tesla, @elon, @Tesla_Optimus

🎥 Introducing SAGE, an agentic system for long video reasoning on entertainment videos—sports, vlogs, & more. It learns when to skim, zoom in, & answer questions directly. On our SAGE-Bench eval, SAGE with a Molmo 2 (8B)-based orchestrator lifts accuracy from 61.8% → 66.1%. 🧵 https://t.co/JkGlw3Ad1Z

Agent demos often fail for reasons that are hard to see: unclear tool traces, silent failures, and changes that improve one behavior but break another. Our new course with @Nvidia shows how to use their NeMo Agent Toolkit to surface these issues with OpenTelemetry tracing, run evals that expose brittle reasoning, and deploy workflows with authentication and rate limiting so they behave consistently in real environments. Taught by Brian McBrayer (@Pr_Brian). Enroll now in Nvidia's NeMo Agent Toolkit: Making Agents Reliable: https://t.co/FvQvUuj5lE

OK I had to record a quick video and share a dialog showing my first few tests: https://t.co/800pSOmNmD Dialog: https://t.co/Vev8lHDndn In the video, I show how easy it can be to train a model on a custom task with your own reward function. LMK what I should try next :)

See how @intelligenceco built Cofounder, an AI chief of staff that turns business documents into agent-ready context at scale. 📄 LlamaParse handles continuous ingestion from @gmail, @SlackHQ, @linear, @notionhq, and @github every 30 minutes - processing PDFs, images, and attachments with agentic OCR 🤖 Two-stage retrieval system combines vector similarity with agent reasoning to filter by time, source, and ownership across multiple business tools 💰 Achieved lower costs and latency compared to managed RAG solutions while avoiding weeks of custom parser development ⚡ Freed engineering time to focus on core differentiator - building agents that can act - instead of document infrastructure "It probably would've taken us a month or more to build a worse document parser ourselves. LlamaParse let us focus on the agents instead of reinventing infrastructure." Read the full case study: https://t.co/1C2qyOAKaz

On the latest Unsupervised Learning, I sat down with @echen. Edwin is the founder and CEO of @HelloSurgeAI, the >$1B revenue infrastructure company behind nearly every major frontier model. Some favorite parts: - Why benchmarks make models worse - Why the model companies are diverging as they optimize for different objectives - The four rare qualities that make elite AI evaluators - The future of the model landscape - Why every company should eventually train their own models Listen here: YouTube: https://t.co/FrCSVWh1v1 Spotify: https://t.co/zgS8MQlYJ5 Apple: https://t.co/qUkHXu7kee

+2 more

The pattern we keep seeing: Easy metrics → hard problems later Hard measurements → real progress The labs winning are the ones willing to do the difficult, expensive work of measuring what actually matters. Full conversation: Youtube - https://t.co/Bv6nvbt5kk Spotify - https://t.co/eL6a3HG8wb Apple - https://t.co/9e7awbFfde

+1 more

Fitness predicts mortality risk better than alcohol consumption. In this study, consuming any alcohol or increasing alcohol intake over 10 years was associated with a 20–25% increase in all-cause mortality (vs. abstinence). But being unfit raised mortality risk by 65%, even among non-drinkers! Fitness also protected against alcohol consumption—fit people who consumed alcohol had the same mortality risk as fit people who abstained.

Excited to announce our latest Gemini 3 family addition --- 3.0 Flash which offers frontier intelligence at scale at a fraction of the cost. ⚡️⚡️⚡️ It pushes the pareto frontier of quality vs cost and speed. https://t.co/qahH8cCRoJ https://t.co/KlW0HMc98d

No signs of an end to rapid gains in AI ability at ever-decreasing costs (which is a log scale) yet. I have to update this monthly or more frequently at this point. All AI benchmarks are flawed, but GPQA Diamond has been a pretty good one, though likely close to being maxed out. https://t.co/jAvDz8OczQ

💡🔍 Horizon is using AI-powered insights to turn complexity into clarity. At #AWS re:Invent, we spoke with the startup about its innovation journey & its backbone of AWS services. Join AWS Activate to accelerate your startup journey. 🚀 https://t.co/AKAbwKchgl https://t.co/yhByGNkn5W

Introducing MARS, your first personal AI robot First units shipping this weekend https://t.co/1X1OHn1xfK https://t.co/eDROMc2Ymw

Introducing MARS, your first personal AI robot First units shipping this weekend https://t.co/1X1OHn1xfK https://t.co/eDROMc2Ymw

@nicolaygerold @AmpCode Yo is this documented somewhere? Or is it arbitrary front matter you are putting in the skill? (Or am I misunderstanding completely?) Thanks for sharing https://t.co/LGEKGNRvM1

The @huggingface librarian-bot learned a new trick! Paper recommender now works as an MCP tool - ask Claude for paper recommendations, and it handles the rest! https://t.co/bRYEaTgHAK

The people of UK want to know the truth. 𝕏 is now the #1 news app in the UK. Trust in censored legacy media is collapsing. https://t.co/zRBYmEzzQD

Anti-mass migration tweet? 18 months in jail Nigerian migrant r*pist? No jail Welcome to the UK. https://t.co/MUMOEVSBNF

A sincere thanks to the team behind Grok and of course, Elon Musk! Italian bureaucracy was a nightmare – until @grok by xAI saved the day! Over several days and hours of patient back-and-forth chats (thousands of words of detailed text!), he guided me step-by-step through registering my documents on the Ministero dell’Interno portal: every field explained, errors troubleshooted, exact attachment formats. He also helped my friend and host complete the complex long-term stay permit (Permesso di Soggiorno) application – translations, critical section tips, and pitfalls to avoid. Grok turned overwhelming red tape into something clear and doable. Best AI ever! Thank you! 🇮🇹📄 A sincere thanks to the team behind Grok and of course, Elon Musk

Meta just released sam-audio https://t.co/ThdN1ZcCgB

🚨This is one of the most disturbing videos of ICE abuse I’ve seen. In Minnesota, ICE agents handcuffed a pregnant woman, forced her onto her stomach, and put their full body weight and knees on her, refusing to let her get up. In the video, you can hear the crowd screaming over and over that she is pregnant. They plead with the agents to stop. The woman is not resisting, not moving, not a threat. It doesn’t matter. The agents ignore them, continue pointing pepper spray at the crowd, and one agent even shoves a woman backward, assaulting her simply for speaking up. People keep begging them to let the pregnant woman off her stomach, to stop putting weight on her body. The agents refuse. The crowd grows louder and more desperate, because they are watching someone who could lose her pregnancy, right in front of them, at the hands of ICE agents. Only after the crowd begins throwing snowballs is the pregnant woman able to get off her stomach. And then it gets worse. The ICE agents begin pepper spraying the crowd. One agent drags the pregnant woman by the arm while she is still helpless and restrained. Pepper spray fills the air, hitting bystanders and the pregnant woman herself as she lies on the ground beneath them. This is not “law enforcement.” This is cruelty. This is reckless endangerment. This is state violence carried out in public, on camera, against a pregnant woman who posed no threat. If this is what they’re willing to do while being filmed, imagine what happens when no one is watching.

VL-JEPA is out! A non-generative vision-language model, based on JEPA. Different from typical data-space autoregressive VLMs, VL-JEPA is trained to predict continuous embeddings in the latent space. (1/N) https://t.co/9lqMJTmqZI