Today's AI Pulse
AT&S pledges up to €2 billion for AI‑focused manufacturing hub in Malaysia
AT&S announced an investment of up to €2 billion to establish a new production facility in Malaysia, targeting the fast‑growing AI market in Southeast Asia. The funding will support advanced PCB manufacturing and AI‑related R&D, positioning the company to capture regional demand for AI hardware.
Also developing:
By the numbers: Waymo acquires Apple’s Arizona Proving Ground for $220M
Visa Unleashes OpenAI Alliance and Programmable Money Rails to Power the AI Agent Economy
At the Visa Payments Forum, Visa unveiled a sweeping architecture upgrade called Visa Intelligent Commerce, designed to let autonomous AI agents initiate and settle payments. The rollout includes a strategic partnership with OpenAI, which will supply natural‑language interfaces while Visa provides the secure transaction backbone. New validation tools – Agent Score and the Agentic Directory – help merchants certify AI‑driven checkout flows and block unverified agents. Visa also introduced a Large Transaction Model that leverages real‑time token assurance to cut false declines and curb sophisticated fraud.
Carney’s Middle Powers Race to Thwart US-China Dominance of AI
Middle powers are mobilizing to prevent the United States and China from monopolizing AI development. Canada, the EU, Japan, South Korea and others are forging alliances, funding AI factories, and acquiring strategic tech to build independent capabilities. The push follows...
The Consumption Gap: How AI Is Splitting SaaS Pricing Into Two Layers
AI shopping agents now account for roughly 20% of holiday ecommerce orders and deliver conversion rates three times higher than traditional checkout flows, prompting platforms like Shopify and Stripe to rethink pricing models. The surge in AI‑driven consumption is creating...
Gaming AI Search
AI‑driven search engines like ChatGPT, Claude and Gemini are surging, growing over 300% while traditional browser search declines. Brands that lack earned media exposure risk being invisible in these agentic searches, as more than 90% of AI‑generated results pull from...
The 6 Best AI Governance Tools in 2026
Zapier’s editorial team evaluated six leading AI‑governance platforms for 2026, highlighting Zapier, Microsoft Purview, OneTrust, Credo AI, Fiddler, and ModelOp. The review breaks governance into workflow control, GRC compliance, and runtime observability, and scores each tool on visibility, policy enforcement, automated...
The 4 Best AI Website Builders
AI website builders promise to create functional sites in minutes, but performance varies. After testing over 50 tools, Wix, Jimdo, Framer, and Chariot emerged as the most capable, each excelling in a different use case. The review highlights how AI...
What Did AI Find Hidden in 35 Years of Hubble Images?
European Space Agency researchers used the AI tool AnomalyMatch to scan 99.6 million Hubble image cutouts, uncovering more than 1,300 visually unusual objects, including over 800 never before documented. The catalog features new gravitational‑lens candidates, jellyfish‑type galaxies and dozens of mergers....
Satya Nadella Says AI Backlash Is Real but Predicts Higher Wages and Broader Prosperity
Satya Nadella told the Hard Fork podcast that AI backlash is real across the United States, but he believes the technology will ultimately lift wages and make everyone a stakeholder in its wealth. He highlighted Microsoft’s $12 billion investment in OpenAI...
Google Promised Never to Build AI Weapons; Then Director René Mayrhofer Resigned when the Tech Giant Broke that Promise
Senior Android security director René Mayrhofer quit Google after the company signed a Pentagon contract that gives the U.S. Department of Defense access to its AI systems for classified military work. Mayrhofer’s farewell letter accuses Google leadership of abandoning its 2018...
Why Gradient Descent Still Fails: Key Preconditions Identified
On a first read, this paper seems far ahead of the pack in terms of (1) understanding some reasons why a task might stay difficult even in the face of gradient descent, and (2) distilling out propositions they'd need to...
OpenAI Acquires Secure‑Cloud Startup Ona to Power Codex Enterprise Rollout
OpenAI announced the acquisition of Ona, a cloud‑execution and orchestration startup, to embed secure, long‑running environments into its Codex AI coding assistant. The move aims to make Codex viable for finance, healthcare and government customers that demand strict data residency...
AI Can Generate Answers but the Future of Expertise Lies Elsewhere
Artificial intelligence now produces technically polished proposals with minimal effort, eroding the traditional marker of expertise—high‑quality output. The article argues that true differentiation will shift from answer generation to framing the right questions, exposing hidden assumptions, and adapting recommendations as...
Jeff Bezos’s Prometheus Raises $12B to Build an ‘Artificial General Engineer’ for the Physical World
Prometheus, the physical‑AI startup co‑founded by Jeff Bezos and former Verily executive Vik Bajaj, announced a $12 billion financing round that lifts its valuation to $41 billion. The round was led by Bezos himself alongside JPMorgan Chase, Goldman Sachs and BlackRock. Prometheus...
.jpg)
Apple’s Camera Chief Thinks AI Can Give You Superpowers
At WWDC Apple unveiled iOS 27’s new generative‑AI tools for the Photos app, including Extend, which adds up to 25% background space, and Spatial Reframe, which shifts perspective while synthesizing missing pixels. The existing Clean Up feature also gains a more accurate AI...
Computer Says Kill: The AI Safety Circus W/ Heidy Khlaaf
In this episode of "Computer Says Kill," AI safety expert Heidi Kloth, chief scientist at the AI Now Institute, explains how the concept of safety in critical systems—originating from the nuclear arms race—has historically driven rigorous risk assessment and reliable...
AWS Launches Graviton5 Instances, Raising the Bar for AI‑powered SaaS Workloads
Amazon Web Services has made its Graviton5‑powered M9g and M9gd EC2 instances generally available, touting up to 25% higher compute performance and 35% faster machine‑learning inference. The move gives SaaS companies a new hardware option to run agentic AI workloads...
Can Artificial Intelligence Replace Your Company’s Editor?
A recent study in the Journal of Writing Research compared professional editors with ChatGPT on four Dutch corporate letters. Human editors consistently improved readability, removed jargon, and avoided factual errors, while the AI’s output varied dramatically based on prompt specificity....
Why We Seek to Fly: Books in Brief
The 2026 "Books in Brief" roundup spotlights four new titles that dissect emerging trends across health tech, biology, and technology philosophy. Robert Wachter’s "A Giant Leap" examines AI‑driven digital scribes and public support for AI in healthcare, while Lorna Gibson’s...
AI Bots Keep Overloading Servers. Should Website Owners Keep Paying? Via @Sejournal, @Martinibuster
AI bots are generating a surge in automated traffic, up 300% year‑over‑year, and are straining website servers far beyond traditional scraping concerns. Roughly 80% of this crawl activity serves large‑language‑model training, often looping through dynamic pages like carts that bypass...
UC San Diego Unveils Next-Gen Data Center to Support AI Research Growth
UC San Diego’s Computer Science and Engineering department opened a 2,900‑square‑foot data center designed as both a high‑performance research platform and a hands‑on learning lab. The facility delivers 1.3 MW of power, 100 Gbps network uplinks (expandable to 400 Gbps), and in‑rack liquid...
Why Is It so Hard to Get ROI From AI? Because Building From First Principles Isn’t Easy
Executives at Fortune Brainstorm Tech warned that many firms are still chasing quick AI wins without laying the essential groundwork. They highlighted the need for foundation work—clean data, robust governance, and re‑engineered workflows—before AI can deliver measurable returns. Leaders from...
Xiaomi's New Open Source, Agentic AI Coding Harness MiMo Code Beats Claude Code at Ultra-Long, 200+ Step Tasks
Xiaomi has open‑sourced MiMi Code V0.1.0, a terminal‑native AI coding assistant that integrates with its MiMo‑V2.5 multimodal model and a million‑token context window. The tool adds a cross‑session SQLite‑FTS5 memory system, checkpoint‑writer subagent, and self‑improvement commands to preserve state over...
Providence Continues to Measure Impact of Ambient AI on Its Clinicians
Providence Health System released a new JAMA Network Open study evaluating its ambient clinical intelligence tool, Dragon Ambient eXperience (DAX). The analysis of EHR metadata from July 2023 to March 2025 showed a sharp drop in average note‑writing time during the first...
Upriver Raises $14M to Automate Enterprise Data Engineering for AI
Upriver Data, an Israeli startup founded in 2024, announced a $14 million seed round led by Valley Capital Partners and Hetz Ventures. The company’s AI‑native platform automates end‑to‑end data engineering, fixing quality issues, maintaining pipelines, and generating datasets without manual effort....
Helix Launches with $10B+ in Funding to Build AI Infrastructure
Helix Digital Infrastructure launched with more than $10 billion in long‑term capital commitments to build AI‑focused data centers for hyperscaler cloud providers. Backers include KKR, the Kuwait Investment Authority, Nvidia and energy utility Vistra, while former AWS CEO Adam Selipsky will...
Claude Is Ready for Its Corporate Close-Up
IDC’s latest report flags Anthropic’s Claude as a rising contender in enterprise AI. While only 19% of firms use Claude heavily, the company has shifted to usage‑based pricing and launched a Claude Partner Network to deepen market penetration. Anthropic projects...
Amazon Adds AI Image Search to Its Shopping App
Amazon announced on June 3 that its Shopping app now includes an AI‑powered image generator directly in the search bar. As users type a description, the tool creates real‑time visual suggestions, letting shoppers pick the closest image and be routed to...
Google Is Held Liable for False Information From Its AI
A Munich court ruled that Google is liable for defamatory statements generated by its own AI overview, ordering the company to remove the false content. The decision rejects Google’s claim that AI‑generated output is beyond its responsibility. Legal experts say...
The Paradox Of The Jevons Paradox: Post AI Law Practice When The Sky’s The Limit
The article argues that AI’s efficiency boost in law triggers a modern Jevons paradox, where lower drafting costs lead to more legal filings rather than fewer. A study by MIT and USC researchers found a 23% rise in docket entries...
Microsoft Copilot Outage Hits Users Again This Month
Microsoft confirmed a second outage of its Microsoft 365 Copilot AI tool on June 11, 2026. The problem originated from a recent deployment, prompting the company to revert to a prior build and restore service within roughly 30 minutes. Downdetector logged 1,878...
Building Robust Foundation Models for Digital Pathology
Researchers published a study in Nature Communications describing a new class of foundation models for digital pathology that can handle the variability of slide preparation, staining, and scanning across institutions. By leveraging self‑supervised and semi‑supervised learning, the models learn rich...
AI Isn’t Accounting’s First Rodeo
AI is the latest disruption accountants face, but the profession remains resilient. AICPA leaders at the Engage conference emphasized AI as a force multiplier rather than a replacement. Their Rise2040 survey of 6,000 accountants shows 80% optimism, yet only 11%...
Premium: Wave After Wave of Demand
NVIDIA is accelerating enterprise adoption of agentic AI by releasing open‑source LLMs and new software layers, while leveraging CUDA‑X to boost data‑intensive workloads. The company is deepening ecosystem partnerships to lock in upstream GPU supply and seed downstream capacity for...
Anthropic’s Claude Fable 5 Plays It Too Safe on Safety, Developers Say
Anthropic released Claude Fable 5, its most capable public model, but users quickly reported that its safety system was blocking benign or legitimate prompts. The model routes flagged queries to the less capable Claude Opus 4.8, which Anthropic says affects about 0.05% of...
AI Data Centers Will Outspend Marshall Plan Fourfold
Tech companies will spend $650B on AI data centers in 2026 alone. For context: the entire Marshall Plan — which rebuilt postwar Europe across multiple nations and years — cost roughly $160B in today's dollars. We are funding civilization-level infrastructure...
Why PNC Is Building Its Own 'AI Factory'
PNC Financial Services Group announced it will build its own AI factory, acquiring data centers and Nvidia GPUs to develop proprietary large and small language models for banking tasks such as fraud detection and call‑center automation. The initiative is designed...
Healthcare And Life Sciences: Turning AI Momentum Into Lasting Value
Healthcare and life‑science firms are accelerating AI adoption faster than anticipated, but regulators and professional groups warn that rapid, ungoverned deployments risk a “trust tax.” Past digital tools like electronic medical records failed due to fragmented data and poor workflow...
How Much AI Is in C3.ai?
Founder‑CEO Thomas Siebel labeled C3.ai’s recent sales "entirely unacceptable" and then bought 6.17 million shares for roughly $69 million. The company reported a sharp FY2026 revenue decline and set FY2027 guidance below the prior year, while cutting headcount dramatically. Siebel returned as...
Ballooning AI Costs Have Canadian Startups Weighing Alternatives
Frontier AI providers such as OpenAI and Anthropic are hiking prices, prompting Ottawa‑based startups to reassess their reliance on these models. Companies like Pluvo, Fellow, Backboard.io, Compose Health and Rewind reported soaring expenses, with some paying seven‑figure annual bills and...
From Predictive Maintenance to Autonomous Ops: The Future of Reliability
Honeywell outlined its roadmap to autonomous asset optimization at the 2026 User Group conference, emphasizing AI‑driven workflows, robust data foundations, and expanded APM capabilities following recent acquisitions of Sundyne and Compressor Controls. The company’s maturity model moves firms from run‑to‑failure...
Veritus CEO Joshua March on Deploying Compliant AI Voice Agents in Financial Services
Veritus, founded in 2025, provides AI‑driven voice, SMS and email agents that are built to meet strict financial‑service regulations such as FDCPA, TCPA, FCRA, GLBA and state rules. CEO Joshua March emphasized that compliance is embedded at every interaction, from...
AI Is Becoming America’s New Therapist—And the Risks Are Growing Fast
AI-powered chatbots are increasingly serving as first‑line emotional support for millions of Americans, filling gaps left by therapist shortages and high costs. These digital companions offer 24/7, low‑cost access, but they lack professional judgment and accountability. The rapid adoption is...
Google DeepMind’s TacticAI Can Predict Football Plays 8 Seconds Before They Happen. Palmeiras Is the First to Use It.
Google DeepMind’s TacticAI uses geometric deep learning to forecast football actions up to eight seconds ahead, turning broadcast video into a dynamic graph of player movements. In a study with Liverpool FC, tactical experts preferred the AI’s recommendations 90% of...
True Positive Weekly #165
The True Positive Weekly #165 newsletter curates eight notable AI developments, ranging from a fresh perspective on data science in the AI era to practical guides on CUDA GPU programming. It highlights breakthroughs such as world‑modeling for physical AI, foundation‑model‑based...
Anthropic Recruits Army to Sell Claude to Nonprofits
Anthropic unveiled Claude Corps, a $150 million initiative that will fund 1,000 one‑year fellowships at $85,000 each to help nonprofits integrate its Claude generative‑AI model. The program, run with education nonprofit CodePath, places fellows in roughly 400 host organizations ranging from...
ACLU Sues After Facial Recognition Falsely Identifies Florida Man As a Child Abductor
The ACLU filed a lawsuit after a Florida man was arrested solely on a facial‑recognition match that identified him with 93% confidence as a child abductor. Police relied on low‑resolution, off‑axis cell‑phone photos rather than original footage to generate the...
3 Things Leaders Need to Know From Microsoft Build 2026
Microsoft Build 2026 shifted its focus from developer‑centric demos to enterprise‑ready AI systems. The event introduced Microsoft IQ’s shared intelligence layer, Frontier Tuning for custom model fine‑tuning, and the Azure‑based Microsoft Agent Platform that unifies building, governing, and scaling agents. New...
IBM, ServiceNow Team to Bring AI to Legacy Enterprise Systems
IBM and ServiceNow announced a joint portfolio aimed at turning legacy enterprise environments into AI‑ready platforms. The collaboration blends IBM’s AI, data, and automation tools with ServiceNow’s workflow‑centric AI platform to modernize applications, enable autonomous IT operations, and strengthen data...
Report: Online Students and Faculty Are Aligned on GenAI Use
A two‑year Oregon State University Ecampus study of over 1,600 online students and faculty reveals that both groups use generative‑AI for similar tasks—brainstorming, proofreading, and summarizing—and share comparable mixed emotions about the technology. While 71% of faculty and 66% of...
OpenAI Reportedly In Talks For $500B Data Center Campus On Federal Land
OpenAI is in negotiations to lease the entire 10‑gigawatt data‑center campus that SB Energy is building on former DOE land in Pike County, Ohio. The development, valued at roughly $500 billion, would give the AI firm control of equipment and responsibility for...