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
AI Agents Are Changing Business: 10 Platforms You Need on Your Radar
The post spotlights eleven AI‑agent platforms—from Google Vertex AI and Microsoft Copilot Studio to AWS Bedrock AgentCore and the newly acquired Manus—showing how they embed into everyday business tools. Each solution lets non‑technical users automate tasks such as data retrieval, report generation, and workflow orchestration. The roundup emphasizes that agents are moving from experimental prototypes to functional digital coworkers. It concludes with a practical playbook: start with a familiar tool, automate one repetitive process, then iterate.
ZFLOW AI Improves B300 Inference with Simulation Tuning
ZFLOW AI announced a simulation‑guided optimization of DeepSeek V4‑Pro on PaleBlueDot AI’s 8×NVIDIA B300 platform, achieving a peak throughput of 826 tokens per second. By disaggregating the prefill‑decode stages, the configuration delivered 1.54× higher throughput and 2–3× lower tail latency...
Microsoft Expands AI Model Menu, Loosening Tie to OpenAI as Enterprise SaaS Shifts
Microsoft announced that more than 10,000 Azure customers have run multiple AI models on its Foundry platform, with usage of Anthropic and OpenAI models doubling quarter‑over‑quarter. The shift signals a broader, less‑dependent AI strategy that could reshape the enterprise SaaS...
Catena Labs Raises $30M and Secures OCC Approval to Pursue National Trust Bank Charter for AI Agents
Catena Labs, founded by Circle co‑founder Sean Neville, closed a $30 million Series A led by Acrew Capital and a16z crypto and received Office of the Comptroller of the Currency approval to seek a national trust bank charter. The funding and regulatory...
AI Tools Lead to ‘Clear Racial Disparities’ in Job Hiring
A recent Financial Times investigation reveals that AI‑driven hiring tools produce clear racial disparities, disadvantaging Black and minority candidates. The analysis, based on multiple employer case studies, shows that algorithmic screening lowers interview callbacks for these groups by up to...
Why 80% of U.S. Startups Are Quietly Switching to Chinese AI Models
A Statrys study finds that roughly 80% of U.S. startups using open‑source AI have migrated to Chinese‑developed models such as Quen, Kim and GLM. The shift is driven by the models' lower hardware requirements, reduced licensing costs, and faster deployment...
AI-Hallucinated Citations Are Creeping Into Papers that Shape Clinical Guidelines, Researchers Warn
An audit of 2.47 million biomedical papers identified 4,046 fabricated references, marking a twelvefold increase since 2023. The fake‑citation rate climbed from roughly four per 10,000 papers in 2023 to 56.9 per 10,000 in early 2026, mirroring the rapid adoption of...
Mecalux Scales up Tech Stack to Drive AI Agents Across Software Suite
Warehouse technology leader Mecalux announced a new high‑performance computing platform to accelerate AI agents across its software suite. The infrastructure will train deep‑learning models and let customers activate configurable intelligent agents for analytics, optimization and decision‑making in warehouses. Mecalux, which...
Training Our AI Agents: Why We Became Our Own First Customers
Heymarket built autonomous AI agents for SMS and messaging support and chose to pilot them on its own customer‑service channel. By routing internal support tickets through the bots, the team gathered real‑world performance data, measuring perfect replies, edited responses, and...
The Vatican Takes on Silicon Valley With an A.I. Warning
Pope Leo XIV released the 42,300‑word encyclical “Magnifica Humanitas,” targeting artificial‑intelligence risks and the concentration of power in Silicon Valley. The document warns that unchecked AI threatens human dignity, the dignity of work, and societal cohesion. While the Vatican’s moral authority...
AI, Autonomous Weapons, and the Pentagon’s $55 Billion Bet on Future War
The Pentagon’s Defense Autonomous Warfare Group (DAWG) is seeking a dramatic budget jump from $225 million in FY 2026 to $55 billion for FY 2027 to fast‑track AI‑enabled drones and autonomous munitions. Lawmakers questioned whether DoD Directive 3000.09, which governs autonomous weapon functions, can keep...
Future-Proof: Transform Your Business with AI or Get Left Behind
Enterprises that ignore artificial‑intelligence risk falling behind as AI reshapes every industry. Analysts forecast global AI spending to top $500 billion by 2027, while early adopters report productivity lifts of 20‑30 percent. Companies are scrambling to build talent pipelines, forge vendor...
AI Thermal Imaging Detects Gray Whales, Unveils Ocean Secrets
Thermal cameras and AI are spotting gray whales before humans can, but the real story is what those detections reveal about the ocean. https://spectrum.ieee.org/whales-ai-thermal-camera-tracking?share_id=9525422
Musk and Altman’s AI Rivalry Reaches Boiling Point as IPO Race Heats Up
Elon Musk’s lawsuit against Sam Altman was dismissed, clearing OpenAI’s path to a potential $1 trillion IPO later this year. Meanwhile, SpaceX announced a $1.75 trillion Nasdaq debut, targeting up to $80 billion in new capital and revealing heavy investment in its xAI...
7AI Launches PLAID ELITE Fully Managed Agentic Security Operations Service
7AI Inc. unveiled PLAID ELITE, a fully managed, AI‑native security operations service that combines autonomous investigation agents with human security engineers. The platform ingests, enriches, triages, investigates and responds to alerts without human intervention, delivering continuous follow‑the‑sun coverage. In its...
Samba TV Enters Identity Wars With Unbiased Data Engine
Samba TV launched Project Gravity, an independent identity‑resolution platform that ingests offline consumer lists and matches them to digital profiles for real‑time ad targeting. The solution pushes matched IDs to more than 40 major ad platforms, allowing brands to upload...
Exclusive: How AI Can Use Blood Biopsies to Make Precision Oncology More Accessible
Mayo Clinic and Stanford Medicine unveiled an AI-driven liquid‑biopsy platform that reads methylation tags in blood to map tumor microenvironments (TMEs) across 17 solid cancers. The two‑stage framework—Spatial Ecotyper for tissue data and Liquid Ecotyper for plasma—discovers nine conserved spatial...
Mitsubishi Electric and Chiba Institute of Technology to Co-Research and Develop Homegrown Physical AI
Mitsubishi Electric has entered a three‑year partnership with the Chiba Institute of Technology to co‑develop "homegrown" physical AI and autonomous robotics. The collaboration will create a co‑creation center targeting humanoid, multi‑legged and drone platforms, with early projects focused on infrastructure...
SoftBank Makes Domestic AI Data Centre Move
SoftBank Corp launched a beta of its AI Data Centre GPU Cloud, targeting a commercial rollout in October. The service combines Nvidia’s GB200 NVL72 rack—featuring 72 Blackwell GPUs and 36 Grace CPUs—with Infrinia AI Cloud OS to deliver LLM development...
AI Extracts Chemistry From Images, Boosting Searchable Data
Some of the most valuable chemistry is not written in text. It lives in figures and schemes. AI-based extraction is starting to interpret chemical structures and reaction pathways directly from images, bringing that hidden information into the searchable evidence base....
Tamnoon Introduces Skill-Based AI Orchestration for Autonomous Cloud Defense
Tamnoon has upgraded its AI engine, Tami, into a skill‑based orchestrator that creates custom remediation workflows for each enterprise. Trained on over 6 million real cloud fixes from 800+ accounts, Tami now offers a Remediation Confidence Score and a Safe Vulnerability...
What Supply Chain Leaders Need to Understand About MCP, A2A, and Graph-Enhanced AI
Supply chain AI is moving from isolated copilots to a coordinated intelligence layer built on Model Context Protocol (MCP), Agent‑to‑Agent (A2A) communication, and graph‑enhanced retrieval‑augmented generation (RAG). MCP supplies persistent operational context, while A2A enables autonomous agents to exchange decisions...
Sam Altman Thinks Using AI in Emails and Slack Is ‘Dehumanising’ – and Revenue Will ‘Take a Bit Longer to...
Sam Altman told Australian business leaders that delegating emails and Slack messages to AI feels dehumanising, prompting him to avoid such use. He cautioned that while AI adoption is accelerating, revenue streams are still nascent and will take longer to...
Gartner: AI Is Reshaping B2B Buying, but Human Sellers Still Close the Confidence Gap
Gartner’s latest surveys reveal that while B2B buyers are increasingly turning to generative AI and digital channels for research, 69% still seek human validation of AI‑generated insights at critical decision points. Buyers consult an average of seven sources, with 45%...
These AI Bots Want to Help Fans Navigate World Cup Host Cities
AI-powered travel assistants are being rolled out in multiple FIFA World Cup host cities to help fans plan trips, find accommodations, and navigate local transit. Frisco, Texas partnered with GuideGeek to launch Frankie, a multilingual bot that pulls data from...
What Readers Found When They Asked Their Chatbots About China
A recent WSJ newsletter asked readers to compare AI chatbot responses to questions about China. Contributors—including a Texas private‑equity investor and a Hollywood screenwriter—noticed noticeable differences between answers generated on the U.S. West Coast and those from other regions. The...
Building the Trust Layer for Agentic Payments with AP2 and Verifiable Intent
Google’s Agent Payments Protocol (AP2) and Mastercard’s Verifiable Intent (VI) are being added to the FIDO Alliance to create a shared trust layer for AI‑driven, autonomous payments. AP2 defines cryptographically signed mandates that capture user‑approved checkout and payment constraints, while...
Open Source DockSec Uses AI to Cut Through Vulnerability Noise in Docker Images
DockSec, an open‑source tool created by Advait Patel, combines existing scanners—Trivy, Hadolint, and Docker Scout—with large language models to filter duplicate CVEs and generate concrete remediation instructions. The LLM, selectable from OpenAI, Anthropic, Google Gemini or a local Ollama instance,...
Cloudvisor Gains Anthropic Reseller Rights, Bringing Claude to Startups on AWS Bedrock
Cloudvisor has secured authorization from Anthropic to resell Claude models through Amazon Bedrock, allowing more than 2,000 startups in 75+ countries to run Claude Opus 4.7, Sonnet 4.6 and Haiku 4.5 on AWS without enterprise‑level procurement. The partnership removes spend...
Hotels Overhaul SEO as AI-Powered Travel Search Captures 37% of Bookings
Hotels are overhauling their SEO strategies after a Boston Consulting Group study found 37% of travelers now rely on AI-enabled platforms to plan trips. Industry leaders like Accor and French startup Custplace warn that traditional search tactics no longer guarantee...
Slowing Things Down.
Pope Leo XIV released a 42,300‑word encyclical urging leaders to curb AI’s disruptive impact and demand public oversight of ethical frameworks. The AI Security Institute, a UK‑funded body, reported safety gaps in top models including Anthropic’s Claude and Google’s Gemini....
European Startup Boom Accelerates as AI, Capital, and Talent Converge
European venture‑backed startups are entering a growth spurt, with AI‑driven firms such as Legora and Lovable posting record revenues and funding, while Paris‑based AMI Labs closed a $1 billion round. The surge reflects deeper AI adoption, larger VC funds and a...
Hyundai Motor Group Accelerates Atlas Humanoid Robot Mass Production for AI‑Driven Factories
Hyundai Motor Group announced that its Metaplant America facility in Georgia will become the lead site for mass‑producing Boston Dynamics' Atlas humanoid robot, while a new Software‑Defined Factory division will integrate AI across its plants. The move adds a dedicated...
Consumers Don’t Want More AI – They Want Retailers to Understand Them
Unlimited Conversions has unveiled its Deterministic Cognitive AI platform, a customer‑intelligence solution designed to replace generic personalization with deep behavioural insight. After four years of pilot work with major retailers, the system claims to map 211 variables per shopper, far...
AI‑Driven Hackers Outpace Patches, Mobile Phishing Beats Email in 2026
Verizon's 2026 Data Breach Investigations Report finds AI‑enhanced hackers now exploit software flaws in 31% of breaches, outpacing the 43‑day median patch cycle. Mobile phishing simulations also reveal click rates 40% higher than traditional email attacks, reshaping corporate defense priorities.
UK and Australia Agree to Work More Closely on AI Security
The United Kingdom and Australia have signed a new memorandum of understanding to deepen cooperation on AI security and safety. The pact links the UK AI Security Institute with Australia’s newly created AI Safety Institute, enabling joint research, shared testing...
Big Four Consulting Has 2 AI Nightmares. KPMG’s Answer to Both Is the Same
KPMG announced a global alliance with Anthropic to embed the Claude large‑language model directly into its Digital Gateway platform, giving every employee across 138 countries access to the AI tool. The partnership makes KPMG Anthropic’s preferred consulting partner for private‑equity...
AI Is Taking Over the Most Cursed Job in the World
A growing cohort of gig workers is being recruited to train generative AI models, a role that industry insiders dub the "most cursed job" because of its low pay, relentless pace, and lack of benefits. Platforms such as Scale AI,...
The Security Assumption Agentic AI Just Broke
A recent red‑team test showed an internal IT‑support agent that could read ServiceNow tickets, SharePoint docs and an employee directory reconstruct a confidential re‑organization plan without any policy violation or mis‑configured permission. The exercise highlights a broader shift: 37% of...
AI Agents Plunged the Tech World Into Chaos. Here’s Exactly How That Happened
Wired's Steven Levy chronicles how the release of Anthropic’s Claude Code and the open‑source OpenClaw platform ignited a rapid proliferation of autonomous AI agents, leading to widespread operational chaos across tech firms. Within weeks, agents were embedded in file management,...
Top AI Field Reporting Apps for Contractors in 2026
Construction firms face tighter schedules, labor shortages and regulatory pressure, driving demand for AI‑powered field reporting apps. These tools combine machine learning, NLP and computer vision to automate voice‑to‑text logs, photo tagging and daily report generation on mobile devices. Leading...
Machine Learning System Design Interview #38 - The Retraining Window Fallacy
In a Netflix staff ML engineer interview, candidates are asked how to handle out‑of‑vocabulary (OOV) items that fall outside a model's 30‑day training window without breaking inference SLAs. The post warns against naive fixes like hard‑coded defaults or expanding the...
Alipay Pushes Deeper Into AI Commerce With New Payment Infrastructure
Alipay unveiled an AI‑native payment infrastructure, branding it the world’s first “AI Wallet” and a new service called “Token Pay.” The platform, aimed at AI agents and developers, has already attracted over 100 million users and processed roughly 300 million transactions since...
Stop Treating AI Governance as a Review Layer. Make It Release Infrastructure
The article argues that AI governance must move from a post‑deployment review layer to an integral part of the release pipeline. Traditional compliance models, which audit static software after it ships, fail because AI models and their data sources evolve...
AI in Special Education: Student Voices, Real Stories, and Practical Strategies with Dr. Gina Tesoriero – Easy EdTech Podcast 372
On May 26, 2026, Monica Burns hosted Dr. Gina Tesoriero on the Easy EdTech Podcast to explore how AI tools like ChatGPT can serve as assistive technology for students with disabilities and neurodiverse learners. The conversation highlighted findings from Tesoriero’s dissertation, which...
From AI Agents to Holographic Patients: The Future of Nursing Education Is Happening at The University of Miami
The University of Miami’s Herbert Business School is overhauling its curriculum to embed artificial intelligence in every course, creating new AI majors, minors, and specialized tracks. It introduced the "Miami Method," a framework that pairs AI‑driven tools with human judgment,...
#216: Google I/O, Musk V. OpenAI Verdict, Andrej Karpathy Joins Anthropic & Meta Layoffs
The episode dives into Google I/O’s rollout of the agentic Gemini era, highlighting Gemini 3.5 Flash, Gemini Spark personal agents, and the new AI‑enhanced search box, while noting the model’s quirks and the broader push toward AI agents. It also...
Why Most AI Agents Disappoint in Production (and What to Fix First)
AI agents look impressive in curated demos but stumble when deployed in real‑world production where data arrives late, APIs time out, and state changes constantly. The article identifies four critical guarantees—freshness, semantics, safe write paths, and lineage—that most enterprise stacks...
Post‑production AI Drives Cinematic Leap, Not Video Generation
Everyone thought AI video models would be the biggest differentiator… But creators using @Seedance + Astra with @TopazLabs are proving the real cinematic leap happens in post-production. The AI video race is evolving fast.
Mythos Sets New Standard for Releasing Dangerous AI
Did Mythos usher in a new era of how to release AI models that are too dangerous to be widely used? My latest for @Nature https://t.co/hkG01TWgMt