Today's AI Pulse
DeepSeek previews V4, its next‑gen open‑source AI model
DeepSeek unveiled a preview of its V4 model on April 24 2026 and pledged to release it under a permissive open‑source license. Early benchmarks indicate V4 outperforms all other open‑source systems on code‑generation tasks, extending China’s reach in the AI race.
Also developing:
By the numbers: Tesla acquires AI‑hardware firm for up to $2.0B
Anthropic's Big Swing, Gemini Everywhere All at Once, and OpenAI's Shape-Shift
Anthropic rolled out four updates in four days, debuting Claude Opus 4.7, a design‑focused Claude Design surface, Claude for Word integration, and Routines for Claude Code automation. Google countered with Gemini 3.1 Flash TTS, which tops the TTS leaderboard, and a native macOS Gemini app built in days using its internal Antigravity platform. OpenAI unveiled a revamped Codex desktop app that can see and control the screen, plus GPT‑Rosalind, a life‑sciences‑specific reasoning model locked behind a vetted enterprise program. The newsletter also warned that AI‑related stock hype, exemplified by Allbirds’ rebrand, may be echoing past tech bubbles.
ENCO Launches AI‑Powered SPECai Enhancements at NAB Show 2026
ENCO announced a suite of AI enhancements to its SPECai ad‑creation platform at the NAB Show 2026, including multi‑voice output, emotional tone selection and a video‑spot plug‑in. The upgrades aim to let broadcast sales teams generate radio and digital ads...
Allbirds Abandons Shoes, Rebrands as NewBird AI with $50 Million Funding
Allbirds announced it will sell its shoe IP to American Exchange Group, rename itself NewBird AI and launch a GPU‑as‑a‑service platform. An unnamed institutional investor is providing a $50 million convertible financing facility, and the stock surged more than 500% on...
Surgical Robotics Beyond NVIDIA GTC
NVIDIA unveiled Open‑H, the largest open surgical‑robotics dataset with over 26,500 task demonstrations and nearly 5 million video‑motion frame pairs collected from nine robot platforms across more than ten institutions. Coupled with the Cosmos‑H world model, the company can generate photorealistic,...
Claude Disrupts Design Industry: Figma & Adobe Stocks Tank
Anthropic unveiled Claude Design, an AI‑driven tool that generates graphics, interactive prototypes, slide decks, and single‑page documents through natural‑language prompts. The platform leverages the new Opus 4.7 model, allowing users to refine outputs via conversation, inline comments, and custom sliders, then...
From AI Shopping to Video, Alibaba Is Making the Investments Analysts Want to See
Alibaba accelerated its AI rollout in April, unveiling a suite of models including the 3‑D world model Happy Oyster and the video‑generation engine HappyHorse. The company’s AI spending jumped to roughly 20 billion yuan ($2.93 billion) in the March quarter, about twice...
Anthropic Withholds Powerful Model, Citing Capability Overhang Risks
Claude Mythos is basically a metal detector for ancient landmines... except the beach is the entire internet. Anthropic's monster new model has been withheld from public release and in this episode of AI from Anywhere, let's talk about why. In...
SiIicon Valley's AI Agent Hiccups: Wasted Tokens and 'Chaotic' Systems
Silicon Valley executives warned that AI agents are still costly and fragile, with many firms squandering millions of tokens on unnecessary model calls. The rise of OpenClaw has sparked hype, but industry leaders from Google, Amazon, Microsoft and Meta highlighted...
Confluence for 4.19.26
Anthropic rolled out three major updates this week: the Opus 4.7 model with adaptive reasoning, Claude for Word in public beta, and Claude Design for AI‑generated visual work. Microsoft’s Copilot now offers Opus alongside its own models, tightening the Anthropic‑Microsoft partnership....
Can Artificial Intelligence Be Governed—Or Will It Govern Us?
The article argues that governing artificial intelligence requires robust institutions, echoing Vannevar Bush’s 1945 vision of a technology‑driven society. Bush’s memex concept foreshadowed today’s internet and highlighted both opportunity and peril, from medical breakthroughs to information overload. The piece stresses...
Innovation Insights Quarterly: Q2 2026
Innovation Insights Quarterly Q2 2026 spotlights breakthroughs in agri‑robotics, biotech, medical devices, water treatment and public‑safety AI. A new AI model enables laser‑weeding robots to identify crops and weeds in real time, cutting retraining from weeks to minutes. Engineered cell therapy...
Dairy Queen Extends AI Drive-Thru Testing Across Franchise Locations
Dairy Queen is expanding its partnership with Presto to deploy voice‑AI in drive‑thrus at dozens of franchise locations, moving beyond earlier corporate tests. The technology aims to boost order accuracy, speed and upsell consistency while easing the chronic staffing challenges...
Why Unlearning Is Vital to Succeed in the AI Era
The post argues that thriving in the AI era requires unlearning entrenched beliefs about work, competence, and decision‑making. It explains how the effort heuristic and presenteeism cause teams to overvalue visible labor, while AI can make people feel smarter yet...
Claude AI Coding: Powerful Now, 100× Faster Soon
I’m spending 2-3 hours a day on Claude coding. It reminds me of my early days of coding when it was so fun to build, but compiling took time, and writing requires many lines of code, … now Claude is...
What's the Deal with Robotaxi Remote Assistants?
The piece examines the emerging role of remote assistants in robotaxi fleets, clarifying that they are not full‑time tele‑drivers but intervene on specific safety‑critical decisions such as school‑bus passes or traffic‑light verification. It outlines a detailed taxonomy that categorises six...
AI Company Known for Teen Suicides Launches New Feature to Turn Books Into Roleplaying Experiences
Character.AI, the chatbot platform criticized for teen suicides, unveiled c.ai Books, a feature that converts public‑domain titles from Project Gutenberg into interactive, choose‑your‑own‑adventure experiences. Users can follow the original plot, diverge in “go off script” mode, or remix stories entirely....
BlackLine, Oracle and OneStream Deploy Autonomous AI Agents as CFOs Boost Budgets Over 50%
BlackLine, Oracle and OneStream are moving autonomous AI agents from pilot to production across global finance platforms in early 2026. The rollout coincides with 25% of chief financial officers planning AI budget hikes of over 50%, signaling a shift from...
MaxLinear Shares Tumble as Analysts Doubt AI‑driven Data‑center Strategy
MaxLinear Inc. saw its shares swing more than 12% lower as Wall Street analysts questioned the company’s AI‑centric data‑center connectivity plan. Despite a 30% revenue jump in 2025 and new 1.6 Tbps products, analysts cite execution risk and a $102 million operating...
ASML Lifts Full‑year Revenue Outlook on Surging AI‑related Equipment Demand
ASML announced a raised full‑year revenue outlook, now forecasting €36‑40 bn for 2026, driven by accelerating demand for AI‑related lithography equipment. Q1 revenue climbed 13% YoY to €8.8 bn, with EUV sales representing two‑thirds of total revenue.
Google Launches A2UI 0.9, a Generative UI Standard for AI Agents
Google announced A2UI 0.9, a generative UI standard that enables AI agents to build and render user interfaces on the fly. The open protocol, backed by an Agent SDK in Python with Go and Kotlin versions forthcoming, aims to cut...
Verisk Analytics Pushes AI‑driven Data Solutions Into Europe’s Insurance Market
Verisk Analytics announced a focused expansion into Europe, targeting Germany, Austria and Switzerland with AI‑enabled data and analytics tools for insurers. The move aims to meet rising demand for precise risk modeling amid tighter regulations and accelerating digital adoption.
Appeals Court Clears Pentagon to Cut Ties with Anthropic in AI Blacklist Dispute
The U.S. Court of Appeals for the Federal Circuit has ruled that the Pentagon may terminate its agreement with Anthropic, ending the company's legal challenge to a “supply chain risk” designation for its Claude AI model. The decision revives a...
Full‑Access AI Needs OS‑Level Sandbox Safeguards
AI agents are far more cable when they have full system access; but when they do, they can mess a lot of stuff up (not unique to any one model). AI harnesses have guardrails: but those can fail. I wonder if...
Indeed CIO Rejects AI Token Leaderboard, Emphasizes Outcome Metrics
Indeed's chief information officer Anthony Moisant told Business Insider the company will never publish a public AI token‑usage leaderboard. He argues that tracking token counts creates perverse incentives and that the firm will instead measure AI success by faster product releases...
Infineon Shares Hit Decade High on Unpriced Price Hikes, Up 81% YoY
Infineon Technologies' shares jumped to a ten‑year peak of €48.30, driven by AI data‑center demand and a price‑increase for power switches that the market has not yet priced in. The rally, which represents an 81% rise over the past year...
Cerebras Systems Files US IPO After Delayed 2024 Offering, Targets Nasdaq Listing
Cerebras Systems, the wafer‑scale AI chip startup, filed a Form S‑1 to list on Nasdaq under the ticker CBRS, marking its return to the market after a postponed 2024 filing. The company, valued at $23 billion after a $1 billion Series H round,...
Why Your AI Assistant Is Suddenly Selling to You
Chatbots are evolving from purely functional assistants to revenue‑generating sales channels as large‑language‑model platforms embed sponsored messages directly into user conversations. The new "sponsored chats" model lets brands pay for placements that appear as natural chatbot replies, targeting users at...
Meta Takes Advantage Plus AI Ads Global, Adding Generative Creative Tools
Meta announced the worldwide rollout of its Advantage Plus AI advertising suite, extending generative creative tools and automated targeting to new markets. The expansion leverages the Andromeda machine‑learning engine, which early data shows can boost ad‑quality metrics by as much...
Microsoft Pushes Copilot Overhaul as Satya Nadella Targets SaaS Growth
Microsoft is reportedly overhauling its Copilot AI assistant, with CEO Satya Nadella directing a push to improve performance and user experience. Analysts note the move comes as Copilot seats sit at 15 million—just 3.3% of the 450 million commercial Microsoft 365 base—while the...
First Token Counts Reveal Opus 4.7 Costs Significantly More than 4.6 Despite Anthropic's Flat Pricing
Anthropic’s latest Opus 4.7 model is priced the same as Opus 4.6 but consumes significantly more tokens per request. Independent measurements show an average token increase of 37 percent, with code‑heavy inputs rising up to 1.47×. For a typical 80‑turn session,...
FAA Opens Bids for AI System to Aid Air Traffic Controllers in $32.5B Modernization
The Federal Aviation Administration announced bids from Palantir, Thales and Air Space Intelligence to build an AI-driven tool called SMART that will warn controllers of potential flight‑path conflicts up to two hours ahead. The effort is part of a $32.5 billion...
Trump-Branded AI Data Center Megaproject Stalls, CEO Departs
The Trump‑branded Fermi America AI data center, billed as the world’s largest, stalled after CEO Toby Neugebauer abruptly departed, sending shares lower following a 75% decline over six months. The project still lacks an anchor hyperscaler tenant and faces cooling‑system...
AI Moves Fast. These Five Things Don't Change.
Product teams are caught in an AI‑focused sprint, tracking every new model from GPT‑5 to Gemini and Claude, yet the core questions of product building remain unchanged. The article argues that fundamentals—problem diagnosis, observing real user behavior, validating solutions, pricing...
Nvidia's Jensen Huang Warns DeepSeek Running on Huawei Chips Is a 'Horrible Outcome' For US
Nvidia CEO Jensen Huang warned that Chinese AI models such as DeepSeek running on Huawei’s domestic chips could undermine U.S. technology standards. He argued that an independent Chinese AI ecosystem would shift global AI standards away from American hardware. Huang...
AI Is Not Your Trading Edge, It Is Your Assistant
AI is reshaping traders' workflow, not creating a new market edge. While AI can accelerate research, backtesting, and data processing, the core insights still come from human interpretation of known technical patterns. The technology excels at surfacing questions about timing,...
AI-Generated Influencers Flood Social Media with Pro-Trump Content Ahead of Midterms
A wave of AI‑generated pro‑Trump influencer accounts has flooded TikTok, Instagram, Facebook and YouTube ahead of the U.S. midterm elections. Researchers identified at least 304 such TikTok accounts since January, with some videos garnering half‑a‑million views and 35,000 followers. Production...
AI Unpredictability Highlights Need for Alignment Over Capability
Yoshua Bengio is highlighting a core challenge in today’s systems. AI can behave unpredictably because it learns patterns without truly understanding intent, making outcomes harder to control in complex situations. The path forward is clear. Improving reliability and alignment is becoming just...
Human Readiness, Not Tech, Drives AI Education Success
AI in education is moving faster than the systems around it. Insights from youth leaders show that success depends less on access to technology and more on human readiness, skills, policies and institutions. The message is clear. Education will not be transformed...
Starbucks Just Launched an AI Order-Picker on ChatGPT. Is It Genius or Insane?
Starbucks has launched a ChatGPT‑powered ordering assistant that lets users tag @Starbucks in the ChatGPT app, describe their mood or even upload a photo, and receive drink recommendations. The AI‑generated suggestions are then transferred to the Starbucks mobile app for...
AI Discovery and Privacy Computing Shift to Operational Investment
AI-driven discovery and privacy-preserving computation are moving from research programs toward operational agendas. Investment choices begin to shift across health systems and digital infrastructure as laboratories connect directly with policy decisions. Microblog @antgrasso https://t.co/UYPQtlJ5Ko
New $10K AI Degree Challenges Harvard with Tech Giants
This CEO has teamed up with Google, Microsoft, and McKinsey to build an AI degree that could rival Harvard—and it will cost only $10,000 to attend https://t.co/t1gWn0rpCv
Taiwan Plays 'Central Role' In AI-Based Computing Transition: Google
Google senior vice president Rick Osterloh said Taiwan is central to the company’s shift toward AI‑based computing, highlighting the island’s R&D strength and partnerships with MediaTek and TSMC. The New Taipei hardware engineering center, Google’s largest outside the U.S., supports a...
AI Video Editing Turns Your Phone Into a Director's Remote
Back in 2012, @mattcohler described great mobile apps as “remote controls for real life”. Thought about that when I used Claude to edit a video … it feels like remote controlling a professional videographer (with a nicer UI than a...
UK Water Utilities Lag AI Leak Detection Adoption
Estelle Brachlianoff is pointing to a missed opportunity. UK water utilities are lagging in using AI to detect leaks, despite the technology being widely applied in more water-stressed regions. It highlights a broader issue. In some sectors, the challenge is not capability,...
'Be Very Good Domain Experts': What Sridhar Vembu Told Zoho Engineers on Surviving the AI Era
Former Zoho co‑founder and ex‑CEO Sridhar Vembu told his engineering teams that deep domain expertise, not just coding skill, will be the key to thriving in the AI era. He cautioned that while AI‑powered coding assistants can accelerate prototype creation,...
AI Complicates Education Rather than Revolutionizing Learning
Until the "AI bubble pops" and "frictionless access to LLMs withers", AI "isn’t revolutionizing education and enhancing learning. It’s just making it extraordinarily difficult to do all the things that have been helping students learn for a very long time"...
Five Essential AI Requirements for Enterprise Agents
Upgrade Your Agent’s Brain: The 5 Non-Negotiables For Enterprise #AI by Rafael Flores @Forbes Learn more: https://t.co/BzGyFiZ6F9 #GenerativeAI #ArtificialIntelligence #MachineLearning #ML https://t.co/TzFv647ONe
D1 Enables Fully Autonomous Campus Last‑Mile Delivery
D1 Is Already Solving Last-Mile Delivery with Fully #Autonomous Campus Runs by @XRoboHub #AI #Robotics #Engineering #Innovation #Technology https://t.co/KSsRZyefIM
Testing a Secret AI Computer—It's Absolutely Wild
I have another computer use AI that is coming that I am testing for a client. Stealth mode right now. This stuff is nuts.
Unified Model: The Future Direction of AI
I have now watched this entire video. It is way more important than I thought it would be. @gravicle lays out a Unified Model and makes a compelling case for why this is the future of AI. https://t.co/AwmyjXmMiZ