Today's AI Pulse
Macron bets on AI with €75 bn SoftBank pledge to power France’s data‑centre push
President Emmanuel Macron is anchoring his legacy on artificial intelligence, securing a SoftBank pledge of up to €75 bn (≈$86 bn) for French AI projects. The initiative, dubbed “Project Marengo,” will use France’s abundant nuclear power to build low‑cost data centres and position the country as Europe’s AI hub.
Also developing:
By the numbers: Waymo acquires Apple’s Arizona Proving Ground for $220M
OpenAI Shifts to Usage-Based Pricing for Codex in ChatGPT Business Plans
OpenAI announced a shift to usage‑based pricing for its Codex model within ChatGPT Business and Enterprise plans, eliminating upfront seat licenses. Administrators can now enable free Codex access across a workspace and pay only for the compute actually consumed, with a limited‑time $500 promotional credit per workspace for eligible customers. OpenAI says over two million developers use Codex weekly and business usage has grown sixfold since January. The move directly targets competitors such as GitHub Copilot and Cursor that still charge per seat.
LeWorldModel Lets AI Simulate Reality for Better Planning
LeWorldModel: Teaching AI to Simulate and Understand the World In this episode of Artificial Intelligence: Papers and Concepts, we explore LeWorldModel, a new approach to building AI systems that can model and simulate real-world environments. Instead of reacting to inputs step-by-step,...
Nvidia and Google Scientists Discuss AI's Next Frontier
What’s next in AI… chief scientists from @Nvidia and @Google talk at #GTC2026. It’s interesting how jittery handoffs were between two esteemed technologists 😊 Bill Dally, Chief Scientist and SVP of Research, @NVIDIA | @JeffDean, Chief Scientist, @GoogleDeepMind and Google Research Interesting...
☕🤖 Last Chance to Join Today's Free Claude Code Masterclass for Beginners
AI Breakers is hosting a free live Claude Code Masterclass today, April 3, at 12:00 PM EDT (5:00 PM WET, 9:00 AM PDT). The session will introduce Claude Code, demonstrate installation, and walk participants through a real‑world task in under five minutes. Attendees...
AI Threatens Gateway Jobs, Stalling Career Mobility
AI may reshape not just jobs, but the pathways between them. Millions of workers without degrees rely on “gateway” roles like admin or customer service to move into higher-paying jobs, yet many of these roles and the transitions they enable are...
Five Claude Features Power 80% of Our Workflow
5 Claude features are doing 80% of the work at my company. I didn't find half of them on my own. My team did. Here's the exact setup we run daily.
AI Breakthroughs, Security Breaches, and Industry Shakeups Define the Week in Tech
Tech week saw a surge in AI investment and model releases, highlighted by OpenAI’s record $122 billion funding round and Alibaba’s Qwen3.5 Omni multimodal system. At the same time, major security incidents—from the NoVoice Android rootkit infecting 2.3 million devices to a...
How to Automate Repetitive Tasks Using Claude Desktop & Computer Use
Claude, developed by Skill Leap AI, now offers autonomous control of macOS computers, allowing the AI to click, type, and navigate applications as a human would. The Dispatch feature extends this capability by letting users trigger tasks from a smartphone, enabling remote...
Supporting MySQL Workflows with Semantic Search with Google Cloud
Google Cloud announced that Cloud SQL for MySQL now natively supports vector indexes, allowing developers to run semantic search and generative AI workloads without a separate vector database. By leveraging Vertex AI embeddings directly inside the database, users can execute...
Aerospike Supports LangGraph Integration with NoSQL Database 8
Aerospike has added LangGraph integration to its NoSQL Database 8, delivering a durable, low‑latency memory layer for stateless AI agentic workflows. The integration provides millisecond response times, high concurrency and fault tolerance, ensuring state persists across crashes, restarts, or pauses. Developers...
Beijing Mandates Internal AI Ethics Reviews to Ensure ‘Controllable’ Tech
Beijing has issued new regulations requiring Chinese companies, universities, and research institutions to establish internal AI ethics review committees effective immediately. The committees must assess AI projects for fairness, controllability, explainability, and personal privacy, extending the 2023 unified ethics review...
"Never Sell" Episode: AI Doom, Veeva, Research and Writing
The latest “Never Sell” podcast episode explores three intertwined themes: the growing investor fascination with AI‑doom scenarios, why biotech‑software leader Veeva Systems appears insulated from disruptive AI forces, and a fresh perspective on investment research that favors a judge‑like stance...
Secure AI Agents Deploy in 60 Seconds, Cut Token Usage.
AI agents are genuinely useful until you see the API bill or realize your ports have been exposed the whole time. @PAIOBot just fixed both problems in one shot sandboxed security built-in, 50% token reduction, deploys in 60 seconds, and free...
Pentagon's Access Demand Leads to First US AI Blacklist
The US government blacklisted an AI company for the first time in American history. Anthropic was already deep inside classified systems. Then the Pentagon demanded unrestricted access. Anthropic said no. Got labelled a national security risk. OpenAI rushed in with their...
WealthStack Roundup: Plancorp Saves 700 Hours a Month Using AI
Plancorp Wealth Management partnered with AI workflow provider CogniCor to test its Advisor Copilot, uncovering more than 700 hours of time saved in a 28‑day pilot. The fee‑only firm, which oversees $8.6 billion in assets, reported that advisors reclaimed 45‑60 minutes...
AI Future: The Leading International AI and Web3 Forum to Take Place in April
AI Future, the flagship event of the global Blockchain Forum, will convene in Moscow on April 14‑15, drawing over 20,000 participants from 100 countries. The two‑day summit features 200 speakers and 250 sponsors, focusing on the intersection of artificial intelligence...
Instagram's Recommendations Outshine AI Investment Hype
Say what you will about the current AI capex megacycle but IG is now delivering me recommended content that goes beyond anything I could have possibly dreamed of
#386 VIRAL VIDEO PROMPT: CHANGE CLOTHES
A Substack post circulates a detailed AI video prompt that instructs generators to keep a man unchanged while a floating outfit swaps instantly onto him. The prompt specifies lighting, shadows, fabric dynamics, and a magnetic‑pull effect for a seamless, cinematic...
Brain‑Network Signal Predicts Depression Therapy Success in New Study
Scientists led by Kaizhong Zheng and Liangjun Chen discovered a brain‑network connectivity pattern that predicts whether patients with major depressive disorder will respond to antidepressants. Analyzing scans from 4,271 participants, their machine‑learning model distinguished future responders with high accuracy, opening...
Is Gitlab Losing Ground in the AI Coding Race?
GitLab’s AI‑driven growth story is under pressure as its revenue outlook dips to a 16% midpoint forecast after a 26% jump last year. The company boasts 50 million users and roughly 10,000 enterprise customers paying over $5,000 annually, a data pool...
Fountain Life Unveils APEX Membership, a Full‑Stack Longevity Program for Biohackers
Fountain Life announced the launch of its APEX Membership, a year‑long, AI‑powered program that bundles advanced imaging, genomics, microbiome analysis and performance testing. The service aims to give biohackers a single, continuous health‑optimization platform.
Claude Code Harness Pattern 2: Tool Architecture and the Tool Contract
The post dissects Claude Code’s Tool architecture, focusing on the comprehensive Tool interface that governs how language models invoke external capabilities. It explains each field—from identity attributes like name and aliases to execution logic, Zod‑based schemas, concurrency safety, and permission...
Scaling AI's Promise in Healthcare: The Time Is Now
ZS CEO Pratap Khedkar warns pharma must move from isolated AI pilots to scalable, high‑impact use cases such as clinical trials, commercialization, and supply‑chain intelligence. He cites a new ZS‑Healthcare Leadership Council report showing the sector is transitioning toward targeted...
U.S. Courts Impose Record $109,700 Sanction on Lawyer for AI‑Generated Filing Errors
A federal court in Oregon ordered a lawyer to pay $109,700 in sanctions for filing AI‑generated citation errors, marking the largest penalty to date. The ruling follows a rapid rise in sanctions—over 1,200 worldwide, about 800 in the U.S.—as courts...
WorkTango Unveils AI‑Powered Coach to Turn Survey Data Into Action
WorkTango announced the launch of WorkTango Coach, an AI‑driven survey analyst that delivers instant insights, personalized action plans and measurable follow‑through for every manager. The product aims to eliminate the weeks‑long analysis bottleneck that has long hampered employee‑engagement programs.
Decentralized Private LLMs Empower User Sovereignty
My self-sovereign / local / private / secure LLM setup, April 2026 by Vitalik Buterin https://vitalik.eth.limo/general/2026/04/02/secure_llms.html
AI Coding Agents Overload CI/CD Pipelines, Raising New DevOps Bottlenecks
Generative AI coding agents are flooding CI/CD pipelines with ten‑times more code, creating integration chaos, review backlogs and hidden technical debt. Engineers report exploding merge‑request queues, divergent credential setups and stalled deployments, prompting a rethink of validation infrastructure.
CuLA: Hand‑crafted CUDA Linear Attention Beats FLA
Alibaba Ant's cuLA (CUDA Linear Attention) The hand-written kernels using CuTe DSL & CUTLASS C++ to extract maximum performance on NVIDIA GPUs. A drop-in replacement for FLA designed to push hardware to its absolute limits. https://github.com/InclusionAI/cuLA
IBM, Arm Team Up on Dual‑Architecture AI Hardware for Enterprises
IBM and Arm have launched a strategic collaboration to create dual‑architecture hardware that blends IBM’s system‑level security with Arm’s power‑efficient designs. The joint effort targets AI and data‑intensive enterprise workloads, promising greater flexibility, reliability and ecosystem reach for CIOs modernizing...
Baidu Robotaxi Outage Halts Over 100 Vehicles in Wuhan, Sparking Traffic Chaos
Baidu’s Apollo Go fleet suffered a system‑wide failure on March 31 that immobilised more than 100 driverless taxis on Wuhan’s ring‑road expressways. Passengers were left stranded for up to two hours, traffic snarled and several minor collisions were reported, prompting fresh...
VOID Erases Objects and Their Physical Interactions in Video
netflix/void-model (open-weight) VOID removes objects from videos along with all interactions they induce on the scene — not just secondary effects like shadows and reflections, but physical interactions like objects falling when a person is removed. https://void-model.github.io/
Elon Musk Unveils Orbital Data Centers, Ties Funding to $75 B SpaceX IPO
Elon Musk announced that SpaceX will launch data‑center satellites to power AI workloads from orbit, and he positioned the plan as a cornerstone of a confidential $75 billion IPO filing that could value SpaceX at $1.75‑$2 trillion. The proposal raises questions about...
AI Finally Closes Search Loop, but Trust Remains Uncertain
The evolution of search has always been "Find". Google never got there but, now, AI has. AI is the first time it feels like the loop is actually closed. But it raises a new question: How much do you trust the answer?
Real Home Robot Maids Are Here: How X Square Robot Merges Automation with Human Partnership
X Square Robot, in partnership with 58.com, has launched China’s first home‑cleaning robot service, allowing consumers in Shenzhen to book a combined human‑robot cleaning team. The AI‑powered humanoid robot autonomously performs structured tasks such as wiping surfaces and organizing items,...
Weekend Reading: AI’s Analysis of Soda and Alcohol Marketing on Social Media
Vital Strategies' Canary platform used AI to analyze soda and alcohol marketing on social media. It found Coca‑Cola appeared in 795 posts during the 2025 FIFA Club World Cup, generating about 6 billion impressions, while digital alcohol ads posted nearly 4,000...
Niantic Turns Pokémon GO Data Into AI Gold
The guys behind Pokémon GO just pulled off the greatest pivot in tech. Shaan and I chatted about this on the pod recently. For years, millions of players walked around cities capturing street-level video data. Now the game's hype is gone. But Niantic is...
Public AI Content Case Studies May Trigger Traffic Drops
Answer: they’re all public case studies about scaling AI content. Usually the case study was written near the time of the peak. Not sure if cause of drop is the content, the case study, or both 😅😅😅
The Hidden Reason Your AI Assistant Feels so Sluggish
AI assistants feel sluggish because agent‑driven workloads overload traditional data platforms. Large language models fire dozens of concurrent short SQL queries per prompt, demanding sub‑second response times that batch‑oriented warehouses can’t meet. This mismatch drives latency and rising costs, prompting...
LLMs Can't Render JavaScript, Trust Their SEO Advice Cautiously
Friendly reminder to be very careful taking technical SEO advice from LLMs; they will confidently make statements about how sites operate that simply aren’t true in many cases, because most LLMs can’t render JavaScript. This includes Claude.
Anthropic Rushes to Contain Leak of AI Coding Tool
Anthropic is trying to limit the damage after sections of its popular AI coding tool were made public. https://t.co/9k3EzyokgX
Ep 748: Plugins, Microsoft’s AI Comeback and New AI Video. 7 New AI Features You Should Be Tracking
In this episode Jordan Wilson breaks down seven fresh AI updates, spotlighting the return of ChatGPT plugins now live in the Codex store, Google Gemini’s new ability to import chat histories, and Slack’s overhaul that transforms its bot into a...
Standard AI Control Plane Will Dominate the Decade
The AI stack today: 🔥 Model layer: crowded, commoditized 🤝 Application layer: fragmented, chaotic ⚡ Harness layer: wide open Whoever builds the standard control plane wins the decade A good AI harness is the next frontier
Anthropic vs Pentagon: AI’s Role in Defense Clarified
The Anthropic and Pentagon standoff has brought AI and the military into the headlines. Here’s what’s really happening. https://t.co/nJ1AThJs3U
Kira Learning Lets You Create Full Courses, Not Just Lessons
📚 Most AI tools help you write one lesson. @Kira_Learning helps you build an entire course. 🗂️ Units, lesson sequences, and assessments all aligned to standards in one place. 🤝 Excited to partner with Kira for this post. https://t.co/5OHA7m8BiF
Agent AI Executes Tasks, Delivers Real Results
Most AI gives you ideas and tells you what to do, but you’re still stuck doing the work and hoping it actually works. Agent AI flips that by taking action itself, handling the execution, and being responsible for getting real...
AI Agents Now Power Real Operations, Not Experiments
Predictive maintenance isn't a pilot project anymore. AI agents are moving into actual operational workflows. Automating decisions, not just recommending them. The organizations still treating AI as an experiment are getting lapped. Are you running or watching?
AI Citations Need Extraction‑friendly, Trustworthy, Machine‑readable Content
Getting cited in AI responses requires more than strong SEO. It demands content built for extraction, trust, and machine readability. https://t.co/uoxw3RU6ay via @slobodanmanic, @sejournal
Chinese Developers Scramble for Anthropic AI Code Leak
This....Anthropic’s AI code leak ignites frenzy among Chinese developers Chinese developers have been having a field day, scrambling to download copies of the leaked code and poring over files to learn detail Forget about distillation.... https://t.co/LhRBnGNaEW via @scmpnews
UBTech Offers $18 M Chief Scientist Salary, Defying Chinese AI Norms
Chinese Humanoid Robot Maker UBTech Offers $18 Million to Hire Chief Scientist The advertised salary is unusual for a Chinese AI industry that’s eschewed the mega pay packages that the likes of Meta Platforms Inc. have put together for elite...
Japanese Firms Adopt Chinese Hospitality Robots Despite Security Concerns
AI startup trains Chinese humanoid robots in Japanese hospitality Japan companies eye tech from China despite security risks due to lack of options https://t.co/bWAquSZSTu via @NikkeiAsia