Today's AI Pulse
Anthropic shuts down Claude Fable 5 and Mythos 5 after US export‑control warning
Anthropic disabled access to its Claude Fable 5 and Claude Mythos 5 models after the Trump administration, via the Commerce Department, warned that foreign nationals could exploit the systems. The move followed alerts from Amazon and other firms about a potential jailbreak that could aid cyber‑security attacks. Anthropic declined the administration’s request for a voluntary shutdown but complied with the order.
Also developing:
By the numbers: AMD acquires AI memory‑optimization pioneer Mext
Adobe Summit 2026 - NBCUniversal Outlines Its Approach to Rollout of AI Agents and Assistants as Adobe Rolls Out New...
At Adobe Summit 2026, Adobe unveiled GenStudio, Brand Intelligence and Firefly‑powered AI agents that can turn a marketing brief into a full campaign in under ten minutes, a task that previously took weeks. The demos showed instant asset retrieval, on‑the‑fly text editing, 3‑D object rotation and automated brand‑compliance checks. NBCUniversal reported over 2,000 creatives using Firefly in production, leveraging custom models to repurpose NBA assets for the Olympics. Adobe also introduced CX Enterprise Coworker, an AI‑orchestrated workflow that can launch a global campaign within minutes of a viral trend.
The Last 1.5mm of AI Power: Three Numbers From Vicor’s Q1 2026 Earnings Call
Vicor Corp. disclosed its second‑generation Vertical Power Delivery (VPD) module on the Q1 2026 earnings call, highlighting a 3 A/mm² current density, up to 40× current multiplication, and a 1.5 mm package thickness. The company also announced a capacity boost at its...
Managing Agents Isn't the Same as Managing Either People or Traditional IT - Digging Into Workday's Agent System of...
Workday’s Agent System of Record (ASOR) introduces a dedicated platform for governing AI agents, distinguishing them from traditional IT assets and human workers. The product defines two core roles—agent admin and agent owner—to manage deployment, guardrails, and usage across the...
AI Is Already Training on Music. The Real Question Is: Who Gets Paid?
AI systems are already being trained on vast libraries of recorded music, often without the creators’ knowledge or consent. While 87% of artists now use AI tools for tasks ranging from mastering to songwriting, the compensation mechanisms lag behind, creating...
The One Thing Apple’s New CEO Needs to Get Right on AI
Apple has appointed hardware veteran John Ternus as its new CEO, signaling a shift toward a hardware‑centric AI approach. Analysts expect the company to focus on running AI models directly on its devices rather than investing heavily in cloud‑based AI...
BioAegis and Prenosis Partner for Inflammatory Disease Therapies
BioAegis Therapeutics has teamed up with AI‑focused startup Prenosis to accelerate precision‑medicine approaches for inflammatory diseases. The partnership will analyze biospecimens from BioAegis’s Phase II BTI‑203 trial of recombinant human plasma gelsolin in 600 ARDS patients across Europe, Canada and the...
You Don’t Need an AI Strategy. You Need a Constraint.
Businesses often feel lost about where AI fits, oscillating between ignoring it and chasing every new tool. The article argues that the missing piece is not a comprehensive AI strategy but a well‑defined constraint that narrows focus. By imposing clear...
Anker Made Its Own Chip to Bring AI to All Its Products
Anker unveiled the Thus processor, a custom AI chip that uses compute‑in‑memory architecture to embed neural‑net inference directly in audio, mobile accessories, and IoT devices. The chip, smaller and lower‑power than conventional AI silicon, will first appear in Soundcore’s upcoming...
The Pope’s Warnings About AI Were AI-Generated, a Detection Tool Claims
A new AI‑detection Chrome extension from Pangram Labs, boasting a 99.98% accuracy rate, flagged a Reddit "Am I the Asshole" post and several Vatican @Pontifex X messages as AI‑generated. The tool scans popular platforms in real time for synthetic text,...
Artificial Intelligence for Inclusive Engineering Education: Advancing Equality, Diversity, and Ethical Leadership
Artificial intelligence is reshaping engineering education through adaptive, data‑driven platforms that aim to close equity gaps. While progress is evident, gender disparity, cultural under‑representation, and limited STEM access persist worldwide. The paper proposes an ethical AI framework linked to UN...
Bond Wants AI to Cure Your Doomscrolling, Then Monetise Your Memories
Bond, a new "post‑feed" social network founded by ex‑Index Ventures partner Dino Becirovic and former Google DeepMind researcher Arthur Brazinskas, launched on April 21 without an infinite scroll or algorithmic timeline. The app lets users upload photos, video and audio...
Prediction of Shear Force in Hanwoo Beef Cuts During Aging Using Advanced Machine Learning
A team of Korean researchers built machine‑learning models to forecast Warner‑Bratzler Shear Force (WBSF) across 33 Hanwoo beef cuts, using a dataset of 15,326 observations from 386 animals. Gradient boosting regression outperformed linear approaches, achieving an R² of 0.64, MSE...
OM in the News: Understanding Manufacturing AI Terminology
Industry Week’s latest guide demystifies the AI buzzwords flooding manufacturing and supply‑chain meetings, from machine learning and large language models to copilots, agents, and embeddings. It outlines concrete use cases—demand forecasting, email summarization, ERP navigation, autonomous inventory actions—and warns that...
Responsible Inference Engines: Safeguarding Students with Learning Differences in the AI Era
A new brief from EALA and New America warns that 73% of students with disabilities are already using AI for coursework, while 57% of special educators rely on it for drafting IEPs. A 2025 systematic review found zero AI‑based interventions...
The Ultimate Guide to Claude Managed Agents 🤖
On April 8 2026 Anthropic unveiled Claude Managed Agents, a set of composable APIs that let developers build and run production AI agents on Anthropic’s cloud without managing runtime infrastructure. The service abstracts sandboxing, session persistence, credential isolation, error recovery and observability,...
OpenAI to Commit up to $1.5bn to PE-Linked Venture
OpenAI is in talks to pour up to $1.5 bn into a new joint venture, DeployCo, backed by leading private‑equity firms. The venture, valued at roughly $10 bn, will see OpenAI contribute an initial $500 m with an option for an additional $1 bn,...
Anthropic Bets on EPSS for the Coming Bug Surge
Anthropic introduced Mythos, an AI model that can discover software flaws at unprecedented speed, intensifying the existing vulnerability overload. To help defenders prioritize, Anthropic advises using the Exploit Prediction Scoring System (EPSS), a probabilistic model that forecasts exploitation likelihood within...
B&H Worldwide Pilots AI Tyre Scanning Across Aerospace Logistics
B&H Worldwide has launched an AI‑driven tyre‑scanning solution at its New Zealand hub, slashing inventory processing time by 60%. The technology, built into the FirstTRAC warehouse management system, uses computer‑vision and OCR on smartphones to capture serial numbers directly from tyre...
Exclusive: OpenAI Briefs Feds and Five Eyes on New Cyber Product
OpenAI has begun briefing U.S. federal agencies, state governments, and Five Eyes allies on its new GPT‑5.4‑Cyber model, a large‑language‑model designed for advanced cybersecurity tasks. The company demonstrated the tool to about 50 cyber‑defense practitioners in Washington, D.C., and announced...
Hands-On with ChatGPT's Powerful New Image Engine
OpenAI rolled out ChatGPT Images 2.0, introducing a new "thinking" mode that adds built‑in reasoning for paid subscribers while keeping a standard version free for all. The update supports a broader set of aspect ratios and can produce personalized cards,...
Top Takeaways for College Leaders From ASU+GSV
The 2026 ASU+GSV Summit in San Diego highlighted how artificial intelligence is reshaping higher education. University leaders discussed AI’s potential to personalize learning for non‑traditional students while warning that technology alone won’t solve systemic challenges. Panels emphasized the need for...
AI Code Saves Time, But Maintenance Remains a Burden
Everyone wants to use AI to automate data work, but nobody wants to review, let alone maintain it. AI-generated code is very similar to written text, except that code is more functional. I did a little self-experiment to test the...
AI Agents Create Unprecedented, Unpredictable Insurance Risk
AI agents present an unprecedented insurance challenge because of their unpredictable behavior, which cannot be compared with business perils such as hurricanes "where the parameters of the threat are easier to define" @FT https://t.co/3mdMpPESBl
EQT Launches AI Infrastructure Strategy
EQT announced a dedicated AI Infrastructure strategy to fund the physical backbone needed for exploding artificial‑intelligence workloads. The firm estimates roughly $4 trillion in global capex will be required over the next five years to expand data centres, power systems and...
Allianz Turkiye Uses Nettle AI to Speed up Risk Engineering Reports
Allianz Turkiye has partnered with Nettle to automate its commercial‑property risk engineering workflow. A pilot showed the AI platform can process inspections up to three times faster, delivering same‑day structured reports, risk scores and underwriting guidance. The mobile app, built for...
General Scales Unlock AI Evaluation with Explanatory and Predictive Power
Researchers led by Lexin Zhou propose a set of "general scales" that transform AI benchmarking from simple scorecards into explanatory and predictive tools. The framework uses 18 cognitive rubrics to map both task demands and model abilities, testing 15 large...
WD's Chief Product Bod Peers Into the AI Data System Future
Western Digital’s chief product officer argues that AI data centers are fundamentally data systems, requiring layered storage architectures rather than pure compute power. He highlights that each inference cycle creates massive logs, traces, and metadata that can rival the size...
My AI Learning Journey – Part 8- OpenRouter – 350+ Models to Experiment With
The post introduces OpenRouter, an aggregation platform that exposes more than 350 large‑language models through a single API key. Unlike traditional subscription services, OpenRouter charges per token, with the author spending roughly $10.85 on experimental projects. Integrated with a self‑hosted...
AI Agents Backed as Long-Term Network Play
Telecom executives at FutureNet World in London agreed AI agents are essential for future network operations but warned that trust, governance, and fault risk limit current closed‑loop use. BT Group’s Reza Rahnama said agents can simplify tool fragmentation and enable...
INTERVIEW: Tiancheng Lou, Founder and CTO, Pony.ai on PonyWorld 2.0
Pony.ai unveiled PonyWorld 2.0, an upgraded world‑model and training system that enhances its autonomous‑driving stack. After confirming profitable unit economics with its seventh‑generation robotaxi fleet in Shanghai and Beijing, the company is accelerating commercialization. It targets more than 3,000 vehicles...
Inside the 2026 AI Frontier Summit: Building, Selling and Leading in the Age of Agents
The 2026 AI Frontier Summit highlighted that most users are still at the earliest stages of AI adoption, forcing founders to simplify interfaces. It emphasized that go‑to‑market repeatability, deep ICP knowledge, and distribution now form the primary moat for AI‑driven...
Enterprise IT Infrastructure Must Enable AI Upskilling, Says Workday UK&I Country Manager
Workday’s UK&I country manager warns that enterprise IT must evolve to support AI‑driven upskilling, as 85% of organisations aim to develop internal talent. Legacy, role‑based HR systems are seen as inadequate by more than half of HR leaders, prompting a...
Ready Teams Don’t Urgently Need AI. The Teams That Need It The Most Aren’t Ready.
The article outlines the "Procurement AI Paradox," where teams with clean data, clear processes and documented criteria can deploy AI agents quickly but already experience low pain, while the most pain‑stricken teams lack the readiness to adopt AI. It argues...
Simpler Is Better for Autograders
Researchers at RAND released a report showing that simple, single‑rubric autograders dramatically reduce evaluation error for open‑ended LLM tasks. Across four expert‑graded benchmarks and five models, the single‑rubric approach cut normalized mean absolute error by 9 to 25 percentage points,...
AI Tool Listens for Endangered Orcas in Real Time to Reduce Human Disturbance
OrcaHello, an AI-driven monitoring system built on the Orcasound hydrophone network, now listens to underwater soundscapes in real time to detect endangered southern resident orcas. The tool has issued alerts on 138 days, prompting ports to pause noisy construction and...
Industrial Disruption and the New Energy Reality - Where AI Makes a Difference
Industrial leaders are confronting persistent geopolitical tension, energy price spikes, and supply‑chain volatility, prompting a shift from reactive tools to agentic AI that can anticipate and act on disruptions. The article argues that traditional record‑keeping systems are insufficient; instead, vertical...
Jeff Bezos’ Project Prometheus Nears $38 Billion Valuation, Targets Industrial AI and ERP-Linked Execution
Jeff Bezos’ Project Prometheus, a "physical AI" lab co‑led with former Google X scientist Vik Bajaj, is nearing a $10 billion funding round that values the venture at roughly $38 billion. Launched in November 2025 with $6.2 billion of initial capital, the startup...
PaleBlueDot AI Launches PBD TokenRouter, a Unified Platform for Accessing AI Models
PaleBlueDot AI unveiled PBD TokenRouter, a unified platform that centralizes access to multiple AI models through a single API. The service offers smart token routing, automatic multi‑channel failover, real‑time cost governance, and intelligent caching to reduce token consumption. Built on...
AI Hype or Reality? Wine Sector Weighs Its Next Move
The Weinheimer Group’s AI Marketing Readiness Report shows 93% of winery owners are experimenting with or researching artificial intelligence, yet only 7% deem it a low priority. Respondents cite online discoverability as the top AI-driven opportunity, while uncertainty and ROI...
Green Building Groups Launch Coalition for Sustainable AI Data Centers
On Earth Day, nine leading built‑environment and sustainable‑finance organisations launched the Greening AI Data Centers Coalition (GADCC) to set transparent, credible standards for sustainable data‑center development as AI‑driven computing demand surges. The coalition, led by groups such as USGBC, BRE...
Master Claude Code in 30 Minutes with Its Creator
🚨 GO FROM BEGINNER TO CLAUDE CODE MASTER IN 30 MINUTES Learning directly from @bcherny (THE creator of Claude Code) teaches you more about vibe-coding than endless hours of tutorials ever could. Inside: → Advanced feats, shortcuts, and workflows power users rely...
Reports: The Know-Your-Customer Agentic AI Revolution; The Definitive Stablecoin Landscape Series; The Shift to Agentic Payments
The latest fintech reports highlight a sweeping shift toward agentic AI‑driven payments, where software agents now initiate, negotiate, and settle transactions across three emerging flows: human‑to‑agent, agent‑to‑business, and agent‑to‑agent. Stablecoins are transitioning from speculative assets to core B2B payment rails,...
AI Doesn’t Just Make You Worse. It Makes You Stop Trying.
A new preprint from Carnegie Mellon, Oxford, MIT and UCLA shows that just ten to fifteen minutes of AI assistance can erode persistence. In three randomized trials with 1,222 participants, those who used AI for direct answers performed worse and...
Artera’s Prostate Biopsy and Breast Cancer Assays Receive CE Mark
Artera announced that its ArteraAI prostate biopsy and breast cancer assays have received CE marking under the EU In Vitro Diagnostic Regulation, expanding the company’s reach to the European market. The prostate assay is the first AI‑enabled prognostic and predictive test...
Autonomous Enterprise: Why 2026 Is the Year AI Agents Accelerate
Enterprise AI is moving from experimental pilots to revenue‑driving agents, spurred by a 2025 MIT study that found 95% of pilots failed to deliver measurable impact. Earnings‑call data shows the word “pilot” dropped 18% in Q4 2025, signaling a shift toward...
AI Chatbots Recommending Chemo Alternatives, Study Warns
A study by the Lundquist Institute evaluated major AI chatbots—including Grok, ChatGPT, Gemini, Meta AI and DeepSeek—on cancer‑treatment queries. Experts rated almost half of the responses as problematic, with 30% deemed somewhat inaccurate and 19.6% highly misleading. The bots often...
Embodied AI Humanoids Reach 310 Units/Hour in China
Watch ‘world-first embodied AI’ humanoid robots take on manufacturing tasks in China 🇨🇳 🤖 💬 System hits 310 units/hour, 19–20 sec cycles, 99 percent success; integrated in 36 hours, producing about 3,000 units per shift. Source: 👉 https://t.co/dQjEBnhTy6 https://t.co/VHcBYyiAN0
GTT Targets AI Networking Growth with 2026 Strategy
GTT Communications unveiled its 2026 strategy centered on AI‑driven networking, expanding its Envision platform and bolstering hybrid‑cloud and security services. The plan builds on 2025’s milestone of over 700 Tbps capacity and reach in more than 170 countries, positioning GTT as...
Not a Chatbot Story. Understanding the Value Proposition of Generative AI in Mining Operations
Generative AI is moving beyond chatbots to address concrete challenges in mining operations. The technology’s value hinges on its integration into decision‑making processes that drive exploration, safety, and equipment reliability. By augmenting human expertise rather than replacing it, AI can...
Calibre Emerges From Stealth with $3.3M to Tackle “Health Guesswork” Through Causal AI
Calibre, a London‑based healthtech startup, emerged from stealth with a $3.3 million pre‑seed round led by Amino Collective. The company introduces “Causal Health Navigation,” a clinician‑guided causal AI platform that identifies the true drivers of an individual’s health. Priced at £69...