Today's AI Pulse
OpenAI unveils ChatGPT Images 2.0 with multilingual text and ‘Thinking’ mode
OpenAI rolled out ChatGPT Images 2.0, a next‑generation image model that adds multilingual typography, real‑time web research, and agentic reasoning. The new ‘Thinking’ mode lets paid users generate up to eight coherent images per prompt, including floor plans, maps and UI mock‑ups. The feature expands the model’s ability to work with uploaded content.
Also developing:
Even Nvidia’s Own Research Teams Can’t Get Enough GPUs Amid the Race for AI Computing Power
Nvidia’s own research groups are scrambling for GPUs, highlighting a sector‑wide shortage of the $30,000‑plus chips that power AI model training. At the HumanX conference, applied‑deep‑learning lead Bryan Catanzaro confirmed that even internal teams must fight for compute allocations. The scarcity has spurred Nvidia to double‑down on efficiency, accelerating development of its open‑source Nemotron family, which is designed to run on fewer GPUs. This internal pressure is reshaping Nvidia’s role from a passive hardware supplier to an active ecosystem steward.
HR Teams Cautiously Experiment with Using AI to Help Set Workers’ Pay
HR departments are cautiously testing AI for compensation decisions, with experimentation rising from 2025 to early 2026 per a Korn Ferry survey. Yet 57% of firms have not started any AI projects in total rewards. Experts warn that data quality,...
Meta Expands CoreWeave Deal to $21 Billion as AI Cloud Demand Grows
Meta has expanded its agreement with AI‑focused cloud provider CoreWeave to a $21 billion, ten‑year contract running through 2032, securing large‑scale GPU clusters for training and inference. The deal comes as Meta ramps up its AI infrastructure spending, targeting more than...
CEO Interview: Axyon AI
Axyon AI targets the $100 trillion-plus pool of institutionally managed assets, offering AI‑driven, human‑in‑the‑loop solutions that automate the entire investment strategy lifecycle. The company estimates a $2.3 billion total addressable market, aiming for 5,000‑7,000 institutional clients with an average annual contract value...
Key Players Pivot to Robots, AI Agents Proliferate | Trading the Markets With AI
The episode covers the rapid expansion of AI agents in crypto (30% of Base chain traffic), Anthropic's Mythos uncovering zero‑day vulnerabilities, and the geopolitical shift toward physical AI and robotics in Japan, China, and the U.S. It highlights Microsoft’s new...
Tubi Strikes ChatGPT Deal that Cuts Out Human Movie Recommendations
Tubi, the ad‑supported streaming service owned by Fox, has launched a new app in the ChatGPT marketplace that lets users ask the chatbot for movie recommendations from Tubi’s catalog. The integration leverages Tubi’s personalization engine, trained on more than one billion...
Seedance 2.0: The Most Controversial AI Video Model in the World Just Landed in the US
ByteDance's Dreamina Seedance 2.0 AI video model, once paused amid controversy, is now live for U.S. users through CapCut. The rollout follows a brief global pause and promises integration across CapCut's Video Studio. CapCut emphasizes safeguards, including limits on real‑face...
Luminai Secures $38M, Inks Enterprise AI Partnership with Cleveland Clinic
Luminai announced a $38 million Series B round led by Peak XV Partners, lifting its total capital to $60 million. Simultaneously, the company unveiled an enterprise AI deployment with the Cleveland Clinic to automate the routing of faxed referrals. By embedding hospital‑specific standard operating procedures...
Cal Newport On Why AI Isn’t Making It Easier
Cal Newport argues that generative AI has not lightened workloads, echoing past tech disruptions. Research from ActivTrak tracking 164,000 workers shows AI users doubled time spent on email, messaging and chat, and increased use of business‑management tools by 94%, while...
Dozens of U.S. Cities Pull Flock Safety License‑Plate Cameras After Privacy Outcry
More than 30 U.S. cities have terminated contracts with Flock Safety, removing its AI‑powered license‑plate readers after public pressure over privacy and data‑sharing with ICE. The wave of cancellations, which began in early 2026, underscores growing scrutiny of municipal surveillance...
Federal Appeals Courts Uphold Pentagon's Blacklist of Anthropic, Escalating AI‑Security Clash
Two federal appellate panels rejected Anthropic's request for a stay, allowing the Department of Defense to keep the AI lab on a supply‑chain risk list. The split rulings—one in Washington, D.C., the other in San Francisco—highlight a growing legal battle...
Automation Frees Us to Pursue New Callings
Two centuries ago, 80% of us worked on a farm. Today, 2%. We didn't become obsolete- we found other callings. @Alfred_Lin brings that long view to Physical AI in Episode 2 of Deployed. The question isn't whether automation replaces tasks. It's...
These AI Glasses Switch Between ChatGPT and Gemini. They're Better Than Meta, but Not Great Yet
Rokid’s AI glasses, now on sale for $279, let users toggle between ChatGPT and Google’s Gemini through a companion app, a flexibility Meta’s $379 Ray‑Ban Gen 2 lacks. Unlike Meta’s glasses, which lock users into a proprietary AI ecosystem and raise...
How Gartner Will Help Accelerate and Scale Your AI Strategy
Enterprise leaders recognize AI’s strategic value, yet many struggle to move from boardroom concepts to production‑grade solutions. Gartner’s Application Innovation & Business Solutions Summit 2026 in Las Vegas addresses cultural, skills, governance, architectural, and cost challenges that impede AI adoption. The three‑day...
Top Influencers’ Secret Weapon for Engagement Might Be Replaced by AI
Influencer marketing has leaned heavily on freelance video clippers who turn long‑form content into bite‑sized TikTok, Reels and Shorts clips. Last year, AI‑powered services such as Overlap and OpusClip began offering end‑to‑end clipping, editing and posting without human involvement. Major...
How 400+ Campuses Are Putting AI to Work
Google's AI for Education Accelerator has enrolled more than 400 colleges across all 50 states in under a year, offering a free AI Professional Certificate that carries ACE‑approved college credit. The program equips students, faculty, and staff with practical AI...
To Prevent AI From Taking Graduates’ Jobs, Comp-Sci Professors Try ... More AI
More than 100,000 U.S. computer‑science graduates are confronting a 7 percent unemployment rate in early 2024, outpacing the 4.4 percent average for all new graduates. To counteract AI‑driven job displacement, professors are weaving artificial‑intelligence projects into senior‑year capstone courses, hoping to turn...
ServiceNow's Context Engine Stops Enterprise AI Stalls
The reason enterprise AI stalls is not the model. ServiceNow built something this week that addresses the actual problem Three announcements landed this morning. Each addresses a different reason enterprise AI programs stall before they deliver returns. Context Engine connects identity, policy,...
Andy Jassy Defends Amazon’s $200B AI Investment in Letter to Shareholders
Amazon CEO Andy Jassy used his 2025 shareholder letter to defend a $200 billion AI capital‑expenditure plan for 2026, emphasizing that the spend is tied to concrete AWS growth opportunities. He highlighted a $4 billion commitment to expand rural delivery networks, positioning...
AI‑Powered Journaling Bridges the Therapy Gap
AI is changing how we work, how we create, how we communicate. It is also starting to change how we know ourselves. This week on TEQ I explore what happens when AI meets journaling, and why the space between therapy sessions...
Top AI Labs Unite on Safety Standards
The Three Biggest AI Labs Just Agreed on Something. That Should Make You Think. https://t.co/6MxkscZbU7
The Threshold Moment
The author recounts a prolonged AI chat that began to lose logical coherence, a phenomenon known as drift. Rather than resetting, they prompted the model to write a blog post about its own breakdown, turning the failure into usable content....
Bus Driver Resists AI Surveillance on School Routes
AI powered cameras are being installed on school buses across the country to surveil the drivers and children on the bus. I spoke to one bus driver in Wichita who’s fighting back against this surveillance tech. https://t.co/IbILY71ldZ
Chinese Open‑Source AI Might Outsmart Anthropic’s Control
Wouldn’t it be something if Chinese open-source models ended up saving the world from oppression by Anthropic? Funniest possible timeline, for sure.
OpenClaw Is Dead — Long Live OpenClaw
Anthropic notified the author that OpenClaw, a Claude‑based personal assistant, would become a metered service, pushing daily costs to $25‑$50. To avoid the expense, the author shifted to OpenRouter, routing calls to the cheaper Kimi 2.5 model, while still retaining Claude...
Control Theory Cuts AI Training Compute Costs
Training a large AI model is expensive, but MIT CSAIL researchers have helped develop a new approach that cuts compute costs. Using control theory, "CompreSSM" compresses models during training, shedding complexity. It makes models leaner & faster: https://t.co/8aGuzy7Mbq https://t.co/yXWqasArmy
Tell AI to Update Its Skill After Every Run
One line that will make every AI skill you build better over time: ↓ After every run, say: "Update the skill with anything you learned this session." That's it. Claude learned from anything that happened in that session and the skill gets...
You’re Looking at the AI Revolution All Wrong, Top Economist Says: 40% Unemployment and a 3-Day Work Week Are the...
Economist Alex Tabarrok argues that AI‑driven productivity could either create 40% unemployment or a three‑day work week, because both scenarios represent the same total labor hours. He stresses that the outcome depends on how society distributes the gains, echoing Keynes’s...
Leverage AI in Trading or Get Left Behind
A lot of people are single-handedly trading without the role of AI. Why would you? Use AI to optimize your trading, backtest, improve and continue to develop with the strongest brain power you could have alongside you. You'll be left behind if you...
Liquid Cooling Unlocks AI-Scale Compute Capacity
Part 2 of the future of the datacenter series. 800 vDC + liquid cooling becomes a co-design situation with compute infra decisions. The key is the compute capacity it unlocks. Liquid Cooling: The Thermal Prerequisite for AI Infrastructure...
3 City AI Policies to Watch
Local governments are rapidly formalizing AI use, with Seattle, San Francisco and Austin each unveiling comprehensive policies. Seattle updated its 2023 framework after nearly 40 pilot projects, tightening bans on bias‑prone applications like emotion analysis and social scoring. San Francisco...
Switching Back to ChatGPT After Claude Proves Unreliable
I’m firing up @ChatGPTapp again. The company and I are relying way too much on @claudeai and it’s no longer reliable or accurate. What a waste of time.
Open PhD Positions in AI-Powered Personalized Nutrition
✨ Several open PhD student positions within the new research platform "AI-powered Personalized Nutrition for well-being and disease prevention" (at Vienna, Austria) 🔗https://t.co/gy3B5DTMf8 🌐 #INPST #DHPSP #Nutraceuticals
Revolut Rolls Out AI Assistant for Daily Financial Tasks
Revolut introduced AIR, an AI‑powered personal assistant, to its UK app on April 9, enabling users to query spending, budgeting, investments and manage card controls through conversational chat. The feature can pause recurring payments, provide live exchange rates, and freeze...
Nvidia's Agent Toolkit Aims to Redefine Enterprise AI Runtime
1/ Interesting: @Nvidia just made a much bigger move up the enterprise stack than most people realize. The new NVIDIA Agent Toolkit not just another AI SDK. It’s a serious attempt to define the runtime layer for enterprise AI agents. If it works,...
Design Products for Agents First, Not as Afterthought
Are you building your product expecting agents to be a primary consumer? Or an afterthought? Good @posthog piece on agent-first product engineering, including some super-solid tips on the right way to build agent skills. https://t.co/sjAGUMtJIq https://t.co/xoEyWtH2QF
Build Your First AI Agent at Social Media Week
At ADWEEK’s Social Media Week on March 15, Code and Theory will run a hands‑on workshop teaching marketers how to build their first AI agents. Attendees will leave with three functional agents—a Trend Pulse Report, a Partnership Evaluator, and an Analytics...
Claude's ROI Forces Me to Demand Lower Vendor Prices
The ROI I get from Claude has broken my brain on all vendor renewals “You want me to pay 2x what we pay for Claude for your little workflow tool?” I have never pushed harder on pricing and threatening to churn in...
Private Grok 4.20 Offers Zero‑Data Retention
The only place to get private Grok 4.20, for video, text, and image. Zero-data retention of prompts and responses. Everywhere else is spying on you.
.jpg?width=1280&auto=webp&quality=80&disable=upscale)
SA Startup Refiant Raises $5M to Advance Sustainable AI
South African‑founded AI startup Refiant secured a $5 million seed round, led by climate‑tech investor VoLo Earth Ventures, to accelerate its nature‑inspired compression technology. The platform claims to shrink massive AI models—such as a 120‑billion‑parameter network—so they can run on a...
Questioning Kimi Claims and Chinese Model Test Validity
Is this true (accurate) about Kimi? If so, is this generally true about Chinese open models? And more importantly, are the tests along this dimension considered to be robust ones? https://t.co/S8SihRyRbI
Athenahealth Targets AI-Driven Patient Discovery with New Rater8 Reputation Management Integration
athenahealth has embedded rater8’s AI‑powered reputation management directly into its athenaOne EHR platform, allowing ambulatory practices to capture reviews, monitor sentiment, and manage directory listings without leaving their clinical workflow. The move is part of athenahealth’s new Alliance Partnerships program,...
Most Health AI Users Don’t Rate Chatbots as Highly Accurate: Poll
A recent Pew poll of over 5,000 U.S. adults shows that while more than 20% occasionally use AI chatbots for health questions, only 18% consider the information very or extremely accurate. By contrast, 65% trust their providers for accurate advice,...
South Africa Chose Flexibility over Control in Its New AI Policy
South Africa released a draft AI policy on April 2, 2026 that favours sector‑specific oversight instead of a single super‑regulator. Existing bodies such as the FSCA, Reserve Bank and SAHPRA will enforce risk‑tiered rules ranging from bans on unacceptable systems...
Speed Up Slide Creation with the Gemini Process
🎨 Creating slides doesn't have to eat up your entire prep period. ✨ I've shared this Gemini process with oodles of educators in workshops, and it's always a hit. 🔗 Now the full step-by-step guide is on the blog. https://t.co/ItAW6EgKbZ
Grayscale Acquired by Paylocity PLUS Indeed's Community Call
This episode of "This Week in RegTech" covers several major shifts in talent acquisition: Indeed’s new pay‑to‑play model that ends free XML feeds and adds daily budget minimums, Oracle’s AI‑driven Fusion Egentic HR applications, Monday.com’s launch of AgentTalent.ai for hiring...
Thousands of Free AI Prompts
An open‑source GitHub repository, “awesome‑chatgpt‑prompts,” has amassed over 155,000 stars and offers thousands of free, role‑based prompts for ChatGPT, Gemini, Claude and other large language models. The collection includes ready‑made prompts for copywriting, Linux terminals, public speaking, debate, social media,...
Lawmakers Are Using AI to Write Laws. What Could Go Wrong?
Lawmakers are increasingly turning to large‑language‑model tools to draft legislation, a shift that began with a 2023 California resolution written by ChatGPT and has accelerated to federal and state agencies adopting AI platforms. Companies such as Vulcan Technologies and FiscalNote...
ENCO Systems Upgrades Its DJ Replacement Technology with Upcoming Release
ENCO Systems pre‑announced an upgraded version of its aiTrack content‑insertion platform, with a formal debut slated for the NAB Show 2026 in Las Vegas. The new aiTrack is a fully cloud‑native, automation‑agnostic solution that leverages large language models, real‑time data...
Claude Mythos Sparks AI Safety vs Innovation Debate
Claude Mythos Preview is the first “too dangerous to ship” moment in AI and this does raise questions about whether AI needs more regulation. I believe it does. But… the balance between providing safety and stalling innovation will make for...