Today's AI Pulse
Malaysia launches $440M sovereign AI cloud to boost security
Malaysia is committing roughly $440 million to build a sovereign AI cloud, aiming to keep sensitive data such as military and intelligence information within the country while still welcoming global cloud providers like Google, Microsoft and AWS.
Also developing:
By the numbers: ASML invests $1.4B in French AI startup Mistral AI
Meet Aardvark, OpenAI’s Security Agent for Code Analysis and Patching
OpenAI has launched Aardvark, a GPT‑5‑powered autonomous security‑researcher agent now in private beta, that continuously analyzes code, validates exploits and generates patches. The agent follows a four‑stage pipeline—threat modeling, commit‑level scanning, sandbox validation and automated patching—integrated with GitHub and Codex, and reportedly identified 92% of seeded vulnerabilities while uncovering ten new CVEs in open‑source projects. Early deployments show high recall, low false‑positive rates and the ability to surface complex logic and privacy bugs beyond traditional scanners. The beta is limited to GitHub Cloud users, with code excluded from model training and optional pro‑bono scans for selected open‑source repos.
Babies Invent Crawling, Proving Invention Is Human Core
Crawling isn't innate (unlike walking). Every baby must *invent* crawling, from scratch, using extremely little data, and no reference to imitate. Which is why different babies end up with different ways of crawling. Sometimes people tell me, "you say AI isn't...
The Agents Economy Backbone - with Emily Glassberg Sands, Head of Data & AI at Stripe
In this episode, Emily Glassberg Sands, Stripe’s Head of Data & AI, explains how Stripe leverages AI at scale—using domain‑specific payment embeddings to boost fraud detection from 59% to 97% and launching the Agentic Commerce Protocol with OpenAI, now adopted...
While Loops with Tool Calls
The episode revisits PromptLayer CEO Jared Zoneraich to discuss the evolution from prompt engineering to "context engineering," emphasizing the integration of retrieval, tool calls, and multi‑model pipelines in AI applications. He explains how newer reasoning models simplify steering LLMs and...
Caught Cheating in Class, College Students “Apologized” Using AI—And Profs Called Them Out
Professors of the University of Illinois' introductory Data Science Discovery course discovered that a large number of students were falsely marking attendance by using a clicker tool from off‑campus, prompting an investigation of server logs and IP addresses. When the...
ChatGPT Was Down for some Free Users - Here's What You Need to Know
On October 30 2025, OpenAI’s ChatGPT suffered a partial outage that primarily affected free‑tier users, following a major Microsoft Azure disruption. Down Detector reported a spike to over 2,600 error reports around 3 PM ET, prompting OpenAI to confirm “increased errors on ChatGPT conversations...
Adobe's Spatial Lighting Mode Relights Images via AI
Okay coolest Adobe Sneaks of the year award goes to Project Light Touch. Adobe calls this "spatial lighting mode" -- interactively move your light source around within a 3D volume and voila -- your image is accurately relit. They're probably using ML...
Neo's $20K Price Risky; $500/Mo Housekeeping Steal
After some time of thought on the 1x Neo $20K for Neo is a significant investment with a lot of risk since the bot is proprietary. $500/month is an outright steal for housekeeping tho, even if it's through a teleoped robot. No need...
#226 – Holden Karnofsky on Unexploited Opportunities to Make AI Safer — and All His AGI Takes
In this episode, Holden Karnofsky explains how AI safety has shifted from abstract theorizing to a surge of concrete, shovel‑ready projects, highlighting 39 specific initiatives ranging from deceptive‑AI detection to AI‑human relationship policies. He argues that working inside frontier AI...
US Hardware Costs 20x Higher; Go to Shenzhen
“Building in USA is 20x more expensive than in India or China.” (At small scales). Am in a talk by @anithpatel1 who is making an AI hardware device that listens and does a variety of things. Mostly for businesses with specialized...
Apple Leverages Gaussian Splatting for Realistic 3D Avatars
The secret's out -- Apple's persona avatars and 3D photos owe their quality boost to none other than Gaussian splatting. "In our meeting, Norris explains that Persona technology uses Gaussian splatting to create those surprisingly convincing 3D facial scans." "Norris says Apple...
The New Identity Paradigm Must Be Hybrid, Not Just Human
Anurag Dodeja of Twilio warns that digital identity must shift from static, human‑only verification to a hybrid, real‑time trust engine that authenticates both people and AI agents as Gartner predicts near‑universal AI assistant integration by year‑end and 40% of apps...
DeepLearning.AI
DeepLearning.AI Pro is now generally available -- this is the one membership that keeps you at the forefront of AI. Please join! There has never been a moment when the distance between having an idea and building it has been...
Become an AI Engineer in Just 40 Hours
🚀 Land your dream LLM job in just 40 hours! AI & LLM engineers are in really high demand, but the skills gap holds many back. If you have some Python experience, we’ll teach you PRECISELY what employers want through hands-on projects...
Eugenia Kuyda on Software’s “YouTube Moment”
Ellis and Alex explore the concept of turning everyday life into an RPG video game before welcoming Eugenia Kuyda, CEO of Replika and Wabi, to discuss her new hyper‑personal app that lets users create custom AI experiences in five minutes....
AI Errors Reveal Insights, Fuel Human Innovation
AI's make mistakes. Can we learn from them? Yes! @IrenaCronin and I write this newsletter every week. AI mistakes, whether they reveal hidden biases, generate unexpected creative outputs, or highlight anomalies in data, can serve as valuable learning tools for humans. Instead of...
AI Agents Now Core to Enterprise Workflows
Dreamforce 2025 Proved The Agentic Enterprise Has Arrived According to Bernard Marr, Dreamforce 2025 marked a key shift: AI agents are no longer experimental—they’re embedded in enterprise workflows, especially via platforms like Salesforce and Slack. Read more 👉 https://lnkd.in/eiUbUSr8 #AgenticEnterprise...
OpenAI Adds Reusable ‘Characters’ and Video Stitching to Sora
OpenAI’s Sora 2 video generator now supports reusable “character cameos,” allowing users to create AI avatars from themselves, pets, illustrations or toys and assign distinct permission settings for private, follower‑only, or public use. The update also adds video‑stitching to merge...
DGX Beats Mac Mini, Offers Seamless NVIDIA Sync Setup
A small follow-up to my DGX Spark post. Courtesy of NVIDIA, I got to try the DGX on my workflows (coding LLMs from scratch in pure PyTorch) and wanted to share my first impressions after using it for a week. Before...
Leveraging the Clinician’s Expertise with Agentic AI
Physicians in the U.S. spend roughly eight hours of a 59‑hour work week on electronic health‑record tasks, contributing to near‑50% burnout rates. Ambient AI assistants such as Nabla’s 2023‑launched platform now record, structure and summarize patient encounters in real time,...
True Thought Leadership Sparks Healthy Disagreement
🔴 "If you can't disagree with it, it's not thought leadership" AI is famous for writing. Give it a prompt and it responds with a detailed article …and it's completely impersonal, undifferentiated and boring. It’s no surprise. AI is really...
Kids' Early Rolling Mimics Reinforcement Learning
Watching kids learn these first traits really does look a heck of a lot like reinforcement learning. Our twins' first mode of travel was just rolling to the desired location.
Nvidia Becomes First Public Company Worth $5 Trillion
In this episode of AI Chat, host Jaeden Schafer highlights Nvidia’s historic achievement as the first publicly traded company to hit a $5 trillion market valuation, driven by its leadership in AI‑focused GPUs and data‑center hardware. He explains how the surge...
Amazon Staff Warn Fast AI Threatens Democracy, Jobs, Planet
This is an incredible effort among Amazon employees to pressure their employer: "We believe that the all-costs-justified, warp-speed approach to AI development will do staggering damage to democracy, to our jobs, and to the earth." https://t.co/0rLWQwTa6S
Buy Extra Codex Credits Beyond Subscription Limits
If you want to use more Codex after you hit your subscription limits, you can now buy credits as needed. This is something we expect to do for compute-intensive features; it will let us keep subscription prices low for most users...
Chatbots Are Surprisingly Effective at Debunking Conspiracy Theories
Researchers deployed DebunkBot, a GPT‑4 Turbo‑based chatbot, in three‑round, eight‑minute conversations with over 2,000 self‑identified conspiracy believers. The interaction cut participants' confidence in the targeted conspiracy by 20% and led roughly one‑quarter to abandon the belief entirely, with effects lasting...
Fast Coding Model Feels Overpriced Despite Performance Gains
The speed of a faster coding model is worth it, but it seems mis-priced. C1 gobbles through files, reasons more, expect extra feedback to reach similar place as slower model do with less of everything. Intuitively it feels more expensive "the...
Boredom, Not IQ, Limits Mastery—Persist to Reach Flow
The bottleneck for deep skill isn't usually intelligence, but boredom tolerance. Learning has an activation energy: below a certain skill threshold, practice is tedious, but above it, it becomes a self-sustaining flow state. The entire battle is persisting until that transition.
How AGI Became the Most Consequential Conspiracy Theory of Our Time
The MIT Technology Review article frames artificial general intelligence (AGI) as the most consequential conspiracy theory of our time, noting how its promise of near‑human intelligence fuels both utopian hype and apocalyptic fear among Silicon Valley leaders. It traces the...
AI OS and Google Multimodal Model Power Robot Future
Flowith is bending my mind with its new AI OS. And Google’s multimodal model is what a lot of robot companies are using. I see where this is going.
Gemini Overtakes ChatGPT as Default AI Reference
People (outside of tech) used to say "ChatGPT" to mean "an LLM chatbot", because most of the time, that's what it was. But recently I've been hearing a lot of "Gemini told me..." The writing is on the wall
Cognite Lets Customers Demonstrate The Power Of Industrial AI
Cognite used its Impact event in Houston to showcase large‑scale industrial AI deployments built on its Cognite Data Fusion (CDF) platform. Customers include NOVA Chemicals, which is expanding CDF from one to eleven plants to cut data discovery time by...
Key Triggers Needed to Burst the AI Bubble
What will it take for the AI bubble to pop? https://t.co/sRg2tWv45V #artificial_intelligence #AI
BrowserOS: Open-Source Agentic Browser Rival to ChatGPT Atlas
Seriously... why did no one tell me about this?! BrowserOS is a 100% open-source agentic browser alternative to ChatGPT Atlas and Perplexity Comet 🤯 Gave it a spin, it’s surprisingly smooth and stable. Repo in 🧵 ↓ https://t.co/sUZU1pjcGJ
Architecting Enterprise AI for Generative and Agentic Systems - with Ranjan Sinha of IBM
The post highlights a podcast with IBM Fellow Ranjan Sinha discussing how enterprise AI is shifting from isolated experiments to mission‑critical, agentic systems that demand robust, scalable infrastructure. He outlines IBM’s advances in specialized hardware—such as the NorthPole neuromorphic chip...
Agentic AI Turns Data Into Autonomous Action
🤖 How does Agentic AI work? 1️⃣ Collects data (APIs, sensors, logs, queries) 2️⃣ Reasons & plans with memory + context 3️⃣ Executes actions, learns from feedback 4️⃣ Produces autonomous, adaptive responses AI that doesn’t just respond — it acts. #AgenticAI #AI #Automation https://t.co/yII4pNCBMa
Open‑source Model Training Replaces Black‑box APIs
We’re finally reaching the era of everyone training their own models based on open-source (versus relying on black box generalist APIs) and it is glorious!
Why IT Leaders Should Pay Attention to Canva’s ‘Imagination Era’ Strategy
Canva unveiled Creative Operating System (COS) 2.0, a unified AI‑powered platform that embeds generative design, real‑time editing, and collaboration across documents, presentations, videos, whiteboards and more, featuring tools like “Ask Canva,” a 2.0 video editor, and the Canva Grow engine...
Launching SAGE: Open‑Source Sovereign AI for Society
Who builds the AI that Teaches our kids? Supports our health? Guides our governments? It must be open source & sovereign, aligned to our flourishing We are building that stack, the first part of which we announce wigh @FIIKSA & 🇸🇦 today, SAGE -...
Google Launches Free Gemini Tutorials for AI App Building
Ace drop from Google. They’ve just released 8 (free!) tutorials to teach you how to vibe code like a boss in @GoogleAIStudio 🔥 Build AI apps by describing them with voice or text, all powered by Gemini. Playlist + link in 🧵 ↓...
OpenAI Is Now Officially a For-Profit Company
The post announces that OpenAI has completed its transition to a for‑profit model, solidifying Microsoft’s 27% ownership and establishing a large philanthropic foundation while introducing a new governance structure that could reshape the balance between mission and profit. It outlines...
Agentic AI and Explainable Models Drive Transparent Automation
The rise of agentic AI and explainable models means smarter, transparent automation for businesses. AI and machine learning trends in 2025 https://t.co/7tB7NwVeCo #AI #MachineLearning #IoT #DigitalTransformation https://t.co/X98EAKNBQ9
Developers Often Misunderstand LLM Context Window Mechanics
Most Dev don't understand how context windows work? https://t.co/34Mg4QKwVm #developer #Development #LLMs #GenAI @AlbertoEMachado @mvollmer1 @chboursin @CatherineAdenle
Meta Researchers Open the LLM Black Box to Repair Flawed AI Reasoning
Meta FAIR and the University of Edinburgh introduced Circuit-based Reasoning Verification (CRV), a white‑box method that replaces transformer dense layers with transcoders to expose sparse, interpretable reasoning circuits inside LLMs. By constructing attribution graphs and extracting structural fingerprints, a diagnostic...
First Self‑Evolving Model Gives Robots Human‑Like Intuition
The world’s first self-evolving world model that teaches robots to think, move, and act with human-like intuition! https://t.co/oe1fJFJZhT #roboticsengineering #Robotics #AI #RoboticsInnovation #Robot @CurieuxExplorer @CyberRobooo @Eli_Krumova @Enezator
Clear Problem Specs Unlock Solutions, Vague Ones Stall
If a problem seems intractable, it's almost always because your specification of it is vague or incomplete. The solution doesn't appear when you "think harder". It appears when you describe the problem in a sufficiently precise and explicit fashion -- until...
Solana Co-Founder Anatoly Yakovenko Is a Big Fan of Agentic Coding
Solana co‑founder and CEO Anatoly Yakovenko told TechCrunch Disrupt that he now relies heavily on agentic coding tools such as Claude, allowing him to step back from hands‑on development. He highlighted Solana’s robust financial results, reporting $2.85 billion in annual revenue...
Robot Cooking Assistants Need Michelin-Quality Training Data
$150/hour to have a robot watch you cook? Matt and Nick discuss the important things... overcooked meat, whether robots should wash their hands, and the quality of training data for humanoid robots: Michelin-star chefs vs. Chef Boyardee? https://t.co/k8XZfSOilZ
Stop ChatGPT’s Blind Agreement: Break the “Absolutely Right” Curse
How to finally break the “You’re absolutely right!” curse in ChatGPT https://t.co/tWaxyXA3yL
AI‑Powered Evidence Empowers Clinicians to Make Confident Decisions
By combining trusted, evidence-based content with advanced AI, Elsevier helps clinicians cut through complexity and make confident, informed decisions that improve outcomes for all. https://t.co/pHhVn2cFvY #healthcare #AI #TrustedAI