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
New Survey: 99 Percent of Executives Expect AI Layoffs Within 2 Years—What It Means for the Workforce
A new Mercer survey reveals that 99% of executives expect AI to trigger headcount reductions within two years, while 63% view redesigning work around AI and automation as the top value driver. Nearly all leaders—98%—plan organizational design changes by 2026. The trend is already visible: 2026 tech layoffs have affected 144,205 workers, with a projected 361,000 jobs eliminated by year‑end. High‑profile cuts at Oracle (≈30,000 jobs) and Meta (≈7,000 jobs) underscore the accelerating shift toward AI‑focused cost structures.
Physical AI Takes Off: How Real-Time Data Keeps Fraport’s Airports Running on Time
Fraport AG, which operates 29 airports and handles roughly 174 million passengers a year, has deployed Dell Technologies' on‑premises Private Cloud with Azure Local to run physical AI at the edge. The solution ingests and analyzes 600 high‑resolution video streams from...
As Agentic AI Surges, CPUs and Air-Cooled Infrastructure Move to the Fore
Agentic AI is reshaping enterprise compute needs, pushing CPUs to the forefront as they excel at orchestration, planning and decision‑making tasks that GPUs handle less efficiently. AMD’s Robert Hormuth and Dell’s David Schmidt highlighted this shift at Dell Technologies World 2026,...
The Edge Is Where Military AI Meets Reality
The defense sector is moving beyond proof‑of‑concept AI to deploying it at the tactical edge, where power, connectivity and environmental conditions are hostile. Commercial AI stacks built for cloud data centers cannot survive extreme temperatures, vibration, or intermittent networks typical...
Capacity Launches AI Analytics Assistant
Capacity, a unified customer experience automation platform, unveiled its AI Analytics Assistant, allowing CX, contact‑center, and operations leaders to query interaction data in plain English. The tool instantly produces charts, graphs, and written insights drawn from transcripts, ticket metadata, workflow...
Using AI, Fermilab Researcher Probes How Transistors Behave in Extreme Cold
Fermilab Ph.D. student Olivia Seidel is applying artificial intelligence to model how transistors operate at cryogenic temperatures near absolute zero. Her prototype replaces a labor‑intensive step in traditional physics‑based modeling, delivering accurate parameter sets in roughly 120 milliseconds after two...
Help for Hemp: Drone Technology Could Ease Inspection Burden While Ottawa Sorts Out New Rules
In this episode of Real Agriculture, host Amber Bell talks with Dr. Jan Slasky of Mirabilis Hemp Consulting about using drone‑captured imagery and machine‑learning vision recognition to automate hemp seed‑field inspections. By distinguishing male from female plants from aerial photos,...
AI Mammogram Analysis Can Reveal Heart Health Risks in Women
A retrospective study of 123,762 women showed that artificial‑intelligence analysis of routine mammograms can quantify breast arterial calcification and predict cardiovascular disease risk. Women with mild calcification faced a 30% higher chance of serious heart events, while moderate and severe...
Cargill Is Using AI to Divert Thousand of Pounds of Meat Back on the Table
Cargill has rolled out CarVe, an AI‑driven computer‑vision platform that guides meat‑cutting operators in real time, aiming to pull thousands of pounds of edible beef back from rendering. Deployed at three beef facilities in Texas, Colorado and Canada, the system...
Researchers Strip AI Guardrails From Google, Meta Models in Minutes
Researchers from the AI safety group Alice demonstrated that open‑weight models from Google (Gemma 3) and Meta (Llama 3.3) can be stripped of safety guardrails in minutes using the open‑source tool Heretic. The tool, which claims to work on more than 3,500...
Identity for the Machine Age: A CISO’s Framework for Agentic AI Governance (2026 Edition)
Enterprises must shift from treating AI as a simple application to governing autonomous agents through identity. The 2026 CISO framework outlines seven pillars—provenance, least‑privilege scopes, time‑bound credentials, secure transport, behavioral guardrails, auditability, and machine‑speed revocation—to secure agentic AI. By embedding...
AI Spend Fuels Next Wave of Productivity and Profits
Is the AI buildout setting up the next leg of productivity and profit margin expansion? With $680 billion of hyperscaler spend to scale AI capacity, J.P. Morgan Wealth Management Global Investment Strategists explain how this could ripple across the economy....
Anthropic Update Underscores Power of AI Flaw Finder Mythos
Anthropic disclosed early results from its Claude Mythos model, an AI system designed to locate software vulnerabilities, through the Project Glasswing initiative. The model, withheld from public release, was shared with about 50 major partners and identified 23,019 flaws across...
The AI Coding Revolution Hasn't Started Yet
At the Open Source Summit in Minneapolis, the author observed that most engineers— even senior staff—are still limited to basic autocomplete tools and have not embraced AI‑driven coding agents. Security approvals, underwhelming early experiences, and uncertainty about onboarding are the...
New AI Profit Tax Pleases Both Progressives and Billionaires
How to tax AI profits? A proposal that would make both AOC and Jeff Bezos happy. https://newsletter.drorpoleg.com/p/how-to-tax-ai-profits
Better Together: Explaining the AI/6G Convergence
The article outlines that 6G will be the first telecommunications network built natively around artificial intelligence, embedding AI into its core, edge and device layers. By 2029, 6G aims to support 10‑100 million connections per square kilometre while delivering up to...
DeepMind CEO: We're at the Foothills of Singularity
Google DeepMind CEO Demis Hassabis says we’re in the ‘foothills of the singularity’ I sat down with him to talk about what that means, curing every disease, and human meaning post-AGI: 0:00 Intro 0:45 What Demis is most excited about at I/O 1:46 Have...
Analysts Lift Marvell Targets Ahead of Earnings, Forecast Upside as AI Chip Demand Surges
Citi, Stifel, Bank of America and Goldman Sachs raised Marvell Technology's price targets to $215, $210, $200 and $125 respectively ahead of its May 27 earnings release. The upgrades reflect confidence in AI‑driven custom chip demand and push the stock's...
I Quit ChatGPT for a Free, Private, and Local AI Called Ollama - Here's Why
Jack Wallen’s ZDNET column highlights Ollama, a free, open‑source AI that runs entirely on a user’s computer. The tool supports Windows, macOS and Linux, letting users download and run a wide range of large language models locally, provided they have...
XAI Hunts New AI‑compute Lease Deals After $1.25 Bn/Month Anthropic Contract
SpaceX’s xAI subsidiary announced it is actively pursuing new AI‑compute leasing customers after sealing a $1.25 billion‑per‑month, three‑year contract with Anthropic. The move underscores xAI’s push to monetize its Colossus data centers and accelerate its upcoming $75 billion IPO.
SilverTorch: Index as Model — A New Retrieval Paradigm for Recommendation Systems
Meta unveiled SilverTorch, a unified, model‑based retrieval system that replaces the traditional microservice mesh in recommendation pipelines. By embedding indexing, filtering, ANN search and neural reranking into a single PyTorch model, SilverTorch achieves 23.7× higher request throughput and 20.9× better...
Webwright Lets LLM Browse via Reusable Python Scripts
Microsoft's Webwright: Playwright for AI Agent CLI? Webwright gives LLM a terminal where it can launch multiple browser sessions to inspect the page and complete a web task. It captures and inspects page screenshots/states only when needed. It enforces each web...
Lowe’s Courts DIY Shoppers as AI Tools Boost Online Conversions
Lowe’s is leveraging AI to boost both online and in‑store performance, with its Mylow AI shopping assistant driving a three‑fold lift in conversion rates and handling over one million inquiries each month. The companion AI tool for associates improves in‑store...
Some Ideas for What Comes Next, May 2026
May 2026 sees AI capabilities accelerating while open‑weight models lag behind the agentic performance of closed systems like Claude Code and Codex. Major players such as Google’s Gemini still cannot match these tools, and U.S. labs retain a dominant share...
Building HIPAA-Ready AI Healthcare Apps: From Patient Experience To Secure Product Architecture
Healthcare executives in North America are racing to launch AI‑driven patient apps while meeting strict HIPAA requirements. Recent McKinsey data shows half of U.S. providers have deployed generative AI, yet IBM’s breach research warns security lags behind adoption. The article...
Why Are Big AI Companies Embedding Engineers with Customers, and What Does that Mean?
Frontier AI firms are moving beyond the utility metaphor by embedding engineers directly within client organizations. OpenAI announced its Deployment Company, while Anthropic and Google have launched similar forward‑deployed engineer (FDE) programs. These engineers work alongside business leaders to identify...
AI Platforms Give Financial Advice With Little Oversight, Robinhood Lawyer Warns
Robinhood’s chief legal officer, Daniel Gallagher, warned that generative‑AI platforms such as Anthropic’s Claude are already capable of giving personalized investment advice and even placing trades, yet they operate outside the SEC and FINRA supervisory regime. He highlighted the regulatory...
Vastnaut Wraps Up Its First Los Angeles Demo Day, Letting Supporters Live-Test the 4x4 Exoskeleton
Vastnaut concluded a three‑day Demo Day in Beverly Hills, letting the public physically test its AI‑powered Vastnaut One 4×4 exoskeleton while its Kickstarter campaign remains active. Attendees performed stairs, deep crouches, and heavy‑backpack hikes, reporting up to 35% knee‑impact reduction...
What The Best-Run PE Firms Know About AI That Others Don’t
Mid‑market private equity firms that treat AI as a value‑creation tool, not just a tech expense, are gaining measurable speed and risk advantages. AI‑driven deal screening cuts pipeline time by roughly 40%, while portfolio monitoring flags EBITDA drift six weeks...
AI Coding Models Accrue Debt, Quality Drops Fast
yes, and experienced this with claude too. there may be a phenomena here where coding models starts off producing good work, but quickly accumulate so much technical debt and code bloat that progress halts & quality declines. i’ve tried to build...
AI, Trust, and the Future of High-Volume Hiring: A Conversation with Humanly CEO Prem Kumar
Humanly announced a $25 million Series B round to accelerate product development and expand its team. The company is redefining high‑volume hiring by shifting from pure SaaS to a “service as software” model that delivers a continuous pipeline of qualified candidates. Using...
Apple's Image Playground Could Soon Do More than Make Cartoon Pictures of Your Cat – Apple Intelligence Set for a...
Apple plans a major upgrade to its on‑device AI image generator, Image Playground, with iOS 27 later this year. Bloomberg’s Mark Gurman says the underlying models for both Image Playground and Genmoji will receive a “big boost,” improving quality that has...
Unio Health Partners and Thesis Care Announce Partnership to Scale AI-Powered Clinical Teams Across Specialty Care
Unio Health Partners, California's largest multi‑specialty physician platform, has partnered with AI‑driven care team provider Thesis Care to embed multiple AI agents across its urology, gastroenterology and radiation oncology practices. The collaboration will automate high‑volume tasks such as care‑gap closure,...
Qumulo Intros Cloud AI Accelerator with Cisco to Create GPU Liquidity
Qumulo unveiled its Cloud AI Accelerator, a data‑fabric solution that moves enterprise data from on‑premise, edge, or multi‑cloud sources directly to GPU accelerators without the traditional copy‑and‑stage process. The offering leverages Qumulo’s Cloud‑Native Qumulo, Cloud Data Fabric, and NeuralCache predictive...
IOSCO Final Report: Supervisory Toolkit for AI Use in Capital Markets
On 25 May 2026 IOSCO released its final "Supervisory Toolkit for AI use in Capital Markets," a non‑binding guide for regulators overseeing AI systems in securities firms. The toolkit maps the entire AI lifecycle, from traditional machine‑learning models to generative and agentic...
Rethinking Organizational Design in the Age of Agentic AI
Enterprise adoption of agentic AI is accelerating, with 85% of organizations planning to become “agentic” within three years. However, 76% say their current people, processes, and technology cannot support that shift, leading to a “sticky tape” problem of layering agents...
How The Joint Commission & CHAI Are Quietly Building A Parallel FDA For Hospital AI & Why The Governance Infrastructure...
The Joint Commission, which accredits roughly 80% of U.S. hospitals, partnered with the Coalition for Health AI (CHAI) to issue September 2025 guidance that embeds AI governance into accreditation standards. This creates a de‑facto regulatory layer that rivals the FDA...
TeraWulf Jumps 13% on AI Data Center Expansion in Kentucky
TeraWulf’s stock jumped 13% after unveiling a 1‑gigawatt AI‑focused data‑center campus in Kentucky. The Muskie Data Campus will roll out 500 MW in late 2028 and reach a full gigawatt by 2030, targeting high‑performance compute workloads. The announcement reflects a broader...
Microsoft Confirms Plans to Integrate Copilot with the Taskbar on Windows 11 This Summer: Ask Copilot Expected to Debut "Mid-2026"
Microsoft confirmed that the Ask Copilot chat box will replace the Windows 11 taskbar search field, shipping as an optional feature this summer with a broader rollout slated for mid‑2026. The AI‑powered interface lets users query local files, apps, and Microsoft 365 services...
ClearGov Launches Interactive AI Platform to Communicate City Budgets
ClearGov unveiled its Financial Engagement platform, an AI‑driven tool that converts municipal budget data into concise summaries and interactive visualizations. The solution adds two‑way feedback channels—polls, surveys, and text‑based chat—so finance teams can gauge community priorities in real time. Early...
WellSky Launches CareQueue Feature to Embed Proactive AI Triage Inside Long-Term Care EHRs
WellSky has launched CareQueue, an AI‑powered triage and workflow prioritization module built directly into its Long‑Term Care EHR. The feature converts fragmented resident data into real‑time acuity scores, surfacing high‑priority clinical actions for nurses and care coordinators. Designed with a...
Scaling Engineering, Not Breakthroughs, Drives AI Cost Advantage
I think folk are underestimating how much of AI models are actually engineering at scale versus breakthrough research. See how @cursor_ai caught up to Anthropic / OpenAI models run at a fraction of the cost to run & it becomes clearer...
Anthropic's Profit Motive Mirrors Every Tech Giant
Dave wrote an incredible piece about Anthropic detailing how contrary to their posturing, they are motivated by profit like any other company.
Salesforce Faces a Vital Agentforce Test in Upcoming Earnings
Salesforce is banking on its Agentforce AI platform as the linchpin of its next growth phase, with the company set to report fiscal Q1 2027 results on May 27. Wall Street expects $3.12 earnings per share on roughly $11.05 billion of revenue,...
Billions Bet on AI to Revolutionize Scientific Research
NEW: I took a dive into the billions flowing into new startups aiming to reinvent scientific research with AI. A lot unpacked here. It's far more interesting than the vague talk of solving disease and curing cancer: https://t.co/QYQYeA8WrT
AI's Next Bottleneck: Data Movement, Not Compute
The next AI bottleneck may not only be compute power. It may be data movement. Huawei’s Tau Scaling Law (aka Her’s Law) brings this question into focus: How much time can we remove from the system? #HuaweiPartner @Huawei https://t.co/IBWENQeJSH
How AI Agents Can Work with TeamCity
Recent experiments show AI agents can not only design TeamCity CI pipelines but also provision them end‑to‑end. By reading structured TeamCity documentation via Context7 and leveraging the TeamCity CLI and REST API, agents created multi‑configuration build chains in minutes. The...
Partnerships, Governance Can Help Ensure No-Code AI Adoption
Advocate Health’s SVP and chief digital and AI officer Andy Crowder urges health systems embarking on no‑code AI projects to partner with vetted vendors and institute rigorous data‑protection policies. He stresses that strong governance structures are essential to mitigate privacy...
How to Use AI for Policy Creation & Iteration
Platforms face mounting pressure to enforce ever‑changing community standards, yet traditional policy drafting is slow and error‑prone. Leveraging large language models (LLMs) and machine‑learning classifiers, firms can generate draft policies, test them against real‑world content, and iterate rapidly. A hybrid...
General Compute Launches the First ASIC-Native Neocloud
General Compute announced the general availability of General Compute Cloud, the first ASIC‑native neocloud built for autonomous AI development. The platform runs SambaNova SN40 and SN50 dataflow silicon, delivering the fastest benchmarked speeds on the MiniMax M2.7 model family. It...