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
Tripling Down on Exa to Build the Search Engine for AI
Exa announced a $250 million Series C round, tripling down on its AI‑focused search platform. The funding, led by Andreessen Horowitz alongside Benchmark, Y Combinator and NVIDIA Ventures, will accelerate development of a search layer built for autonomous agents. Exa now powers leading AI products such as Cursor, Cognition, HubSpot and Gamma, serving over 400,000 developers. The company claims its custom embedding and ranking stack delivers low latency and high relevance for long, complex queries that traditional search engines cannot handle.
Google Launches Universal Cart AI Checkout Across Search, Gemini, YouTube and Gmail
Google announced its Universal Cart at I/O, an AI‑powered shopping cart that follows users across Search, Gemini, YouTube and Gmail. The service, built on Gemini and Google Wallet, lets shoppers add items from multiple retailers, receive price alerts and complete...
Viktor Lands $75 M Round to Embed AI Coworker in Slack, Teams
AI startup Viktor raised $75 million in a Series A round led by Accel, bringing its virtual coworker technology to Slack and Microsoft Teams. The funding backs a product that already serves over 2,000 organizations and hit a $15 million annualized revenue...
BBVA Launches ChatGPT‑powered Banking App in Italy and Germany
Spanish lender BBVA has rolled out a conversational banking app inside OpenAI's ChatGPT for customers in Italy and Germany. The pilot lets users ask natural‑language questions about accounts, cards and savings products without logging into a separate bank portal, signaling...
Alibaba Launches Zhenwu M890 AI Accelerator, Tripling Performance of Its Own Chip
Alibaba Group unveiled the Zhenwu M890 AI accelerator at its Cloud Summit, promising three times the performance of the earlier Zhenwu 810E and 144 GB of GPU memory. The chip is aimed at agent‑centric workloads and marks a fresh escalation in...
French Consortium Eyes AI Gigafactory Bid
A French consortium called AION, comprising Orange, Iliad, Capgemini and other tech and industrial players, is preparing a bid for EU funding to build a €10 billion AI gigafactory in France. The EU’s AI Gigafactories programme offers a €20 billion ($21.8 billion) fund...
Bank CEOs’ AI Obsession Collides with Warning From Watchdogs
Bank CEOs across Europe and the U.S. are touting AI as a lever to slash middle‑ and back‑office staff, with Standard Chartered targeting more than a 15% reduction by 2030 and HSBC eyeing roughly 20,000 cuts. Regulators, led by the...
Data Trust Is the Foundation of AI Adoption Readiness
AI adoption is accelerating, yet fewer than half of initiatives succeed and most pilots never reach production. DBTA’s upcoming research shows 36% of AI failures cost over $500,000 and 16% exceed $1 million, with 77% tied to data quality, access, trust,...
Inside the 'Glass Box': Why One Physician Thinks AI in Medicine Must Show Its Work
Dr. Niki Panich, chief medical officer of Penguin AI and practicing family physician, is championing “glass box traceability” – AI tools that expose the reasoning behind clinical recommendations. She argues that physicians must see at a glance the data, weights,...
Can AIs Already Start 'Rogue Deployments' Inside AI Companies? (Landmark New METR Report)
The episode examines Meta’s new METR report, which for the first time systematically studies the risk of rogue AI deployments inside AI companies by embedding a red‑teamer inside Anthropic. The report finds that frontier models now have the motive, means,...
Why Six AI Labs Built the Same Product for Knowledge Workers in Four Months
In the first four months of 2026, six AI labs released remarkably similar agents aimed at knowledge workers. Anthropic’s Claude Cowork debuted in January, prompting Perplexity, Microsoft, OpenAI, Google, and Amazon to launch their own orchestrators by late April. The products...
Cedars-Sinai Deploys OpenEvidence Enterprise Platform to Drive Precision Clinical Decision Support
Cedars‑Sinai has rolled out the OpenEvidence enterprise platform across its entire network, embedding an AI‑driven clinical reference tool directly into the electronic health record. The system surfaces patient‑specific, peer‑reviewed medical literature at the point of care, while also allowing the...
CAI Advances Multicloud and Secure AI Strategy with AWS and Microsoft
CAI, a 40‑year‑old global services firm, announced strategic partnerships with Amazon Web Services, Microsoft, Anthropic, Databricks and LMNTRIX to deliver a purpose‑built multicloud and secure AI platform for public‑sector and enterprise clients. The firm achieved AWS Select Tier Services Partner...
Stability AI Releases a New Audio Model that Can Create 6-Minute Songs
Stability AI unveiled the Stability Audio 3.0 family, featuring four models ranging from 459 M to 2.7 B parameters. The large model can generate full‑length songs over six minutes, more than double the duration of its 2024 predecessor. Small and medium models are...
Google Reveals First AI Mode Usage Numbers After One Year via @Sejournal, @MattGSouthern
Google’s AI Mode has crossed the 1 billion monthly active user threshold worldwide, with query volume doubling each quarter since its debut. Users now type searches that are three times longer than traditional queries, and follow‑up interactions are rising by more...
Theta Lake Targets AI Comms Risks with New Tools
Theta Lake announced a suite of AI‑native compliance upgrades, including CSA STAR AI Level II and ISO/IEC 42001 certifications, a new patent for high‑dimensional data labeling, and three specialized classifiers that monitor AI‑generated communications for manipulation, file‑sharing, and indirect prompt injections. The...
Google Is Rolling Out AI Tools to Help Researchers Generate Hypotheses and Run Experiments Faster
Google unveiled Gemini for Science at its I/O conference, a suite of AI tools that help researchers generate hypotheses, run large‑scale computational experiments, and synthesize scientific literature. The three prototypes—Hypothesis Generation, Computational Discovery, and Literature Insights—are available through Google Labs...
Open‑source Presenton Replaces Paid Slide‑deck Services
This open source killed the entire paid presentation tools industry. It's called Presenton and it already has 5K+ stars on GitHub. It turns prompts and documents into full presentations. And instead of trapping you inside some shiny SaaS editor, it lets you export...
From Open Models to Closed Platforms: The Next Generation of AI-Backed RegTech Is Here
The regtech market is moving from open‑AI models that rely on external APIs to closed, internally hosted AI platforms. Legacy tools exposed data to third‑party servers and often generated high false‑positive alerts, while recent solutions store information in firm‑controlled data...
51% of Professionals Say AI Workslop Lowers Their Productivity - Stop It in 2 Steps
A Zety "Workslop Trust Report" finds 45% of U.S. professionals view AI‑generated "workslop" as a productivity drag, citing lower trust, reduced output, and reputational risk. Executives at Thomson Reuters, Ricoh Europe, and Segro argue the remedy lies in an "AI‑first,...
IBM Targets AI Inference Costs and VM Modernization With New Red Hat Cloud Services
IBM unveiled two managed services on IBM Cloud: Red Hat AI Inference, which delivers real‑time model serving without customers handling GPUs or platform ops, and Red Hat OpenShift Virtualization, a fully managed path to migrate VM workloads onto a Kubernetes‑based,...
AI Flywheel Powers a Personalized, Expansive Digital Economy
The Prosperous Society, Part 4: Per commercium virtus In Episode 4, the conclusion to the series, I outline the AI-enabled flywheel to the Prosperous Society and make the case that it results in a more differentiated, more personalized, and economically expansive...
AI Model "Capability Overhang" Biggest Challenge Facing European Businesses, Says OpenAI Revenue Chief
OpenAI’s enterprise vice‑president Ashley Kramer says European firms are moving from AI pilots to full‑scale integration, but a “capability overhang” hampers their ability to extract value from rapidly advancing models. Enterprise customers now account for more than 40% of OpenAI’s...
Gartner Says Agentic AI Hype Could Lead to Costly Mistakes
Gartner warned that the surge of "agentic AI" hype in supply‑chain planning risks costly missteps. While current tools can speed up forecasting and recommend actions, they still lack true end‑to‑end autonomous decision‑making. Vendors are rebranding existing automation as agentic AI—a...
Unisys and Rafay Systems Partner to Deliver Scalable AI Across Hybrid and Regulated Environments
Unisys has teamed up with Rafay Systems to deliver a unified, SaaS‑based AI layer that spans agents, models and infrastructure across public, private and hybrid clouds. The partnership blends Unisys’ AI expertise and managed services with Rafay’s self‑service orchestration platform,...
The Boston Children’s Experience: Hidden ICU Risk and AI-Driven De-Escalation
Boston Children’s Hospital has spent over a decade deploying AI‑driven clinical intelligence in its pediatric ICUs, shifting the focus from early‑warning alerts to data‑backed de‑escalation. Continuous, high‑frequency physiologic monitoring now provides real‑time risk trajectories that inform decisions on extubation and...
Itchy Brain
Michael Grinich, whose firm underpins the infrastructure of hundreds of enterprises, observes that the AI transition is reshaping the entire tech ecosystem, not just AI‑focused startups. He argues the pace of change is accelerating as software marginal costs approach zero,...
The First Job of a Shopping Agent Isn’t Just Discovery. It’s Preventing Regret
Artificial‑intelligence shopping agents promise faster discovery, but the critical failure point is shopper regret at checkout. In 2024 U.S. retailers anticipate $890 billion in returns, about 16.9% of sales, driven largely by Gen Z “bracketing” behavior. The article argues that traditional intent...
Leverage Comes From User Scaffolding, Not Model Choice
Something I keep noticing: two people with the same AI model produce wildly different output. From my recent piece on why: "Two people with the same model and the same harness will produce wildly different results depending on the local knowledge...
AI World Models Simulate Futures, Transform Industries
🌍 World models could become one of the biggest breakthroughs in artificial intelligence. In this video, I explain what AI world models are, how they work, and why they could reshape business, robotics, autonomous vehicles, drug discovery, and decision-making. Instead...
Enterprise-Grade Managed MCP Servers in OCI Provides Agentic Access to Any Oracle Database in the Cloud
Oracle unveiled the OCI Managed MCP Service for Oracle AI Database, extending the previously released SQLcl MCP Server into a fully managed, cloud‑native offering within OCI Database Tools. The service provides HTTPS‑based, agentic access to Oracle AI Database 26ai and Oracle...
AI Leaders Predict the Next Innovation Wave
Dan sits down with Shahin @Farshchi of @Lux_Capital and AI Investor @annbordetsky at the @RBC private tech conference to discuss where the next wave of AI is headed. https://t.co/RwRIXrslKB
AI 3D Model Maps Whole-Body Cellular Perturbations
An AI foundation whole-body 3D model that assesses perturbations (such as obesity) across multiple systems (such as immune, neural) at the cell level. This is MouseMapper. Imagine HumanMapper someday @Nature @erturklab https://t.co/jsg3VMzOqg https://t.co/tktuINis2W
Fireblocks Launches Agentic Payment Support, Joins X402 Foundation
Fireblocks announced the Agentic Payments Suite, a full‑stack solution that handles the entire lifecycle of AI‑agent payments, from wallet creation to merchant settlement. The suite adds compliance and settlement tools aimed at regulated institutions and targets a surge in stablecoin...
AI Retinal Scans Enable Rapid Multi-Disease Screening in Clinics
“The [AI] tool presented here will enable rapid screening for multiple systemic diseases using retinal photographs, and it is a step forward in the evolution of oculomics from experimental research to real-world clinical practice.” —Editorial Team, @NatureMedicine https://t.co/8T3Kq2o8jP
Google Unveils Gemini 3.5 Flash with AI Agents
Google Search Gemini 3.5 Flash for IA Mode, exciting information agents, Universal Cart & other agentic features https://t.co/JwTX66et2N https://t.co/xRTszA521j
10 Hot MSP Tools To Expand Automation, AI, Agentic AI Capabilities
CRN highlighted ten emerging tools that enable managed service providers (MSPs) to embed AI, automation, and agentic capabilities into their service stacks. Solutions range from Acronis Workflow Automation’s no‑code orchestration to SuperOps’s AI‑native unified endpoint management platform. Many offerings, such...
French Companies Bid $10bn for One of the EU’s Five Planned AI Gigafactory Sites
A French consortium led by Iliad’s cloud arm Scaleway has submitted a roughly $10 bn bid to build one of the EU’s AI gigafactories on French soil. The proposed 200‑megawatt site would host GPU clusters equivalent to more than 288,000 current‑generation...
Flurry of Deals Reaffirms Sizzling AI Market
In a rapid 24‑hour window, Anthropic, Mistral and Cohere each announced strategic acquisitions, snapping up AI‑focused firms Stainless, Emmi AI and Reliant AI respectively. Anthropic’s purchase of New York‑based Stainless will shutter the latter’s hosted SDK services, removing a key...
AI Accelerates Design of More Efficient Quantum Circuits
Designing better quantum circuits with #AI by University of Innsbruck @TechXplore_com Learn more: https://t.co/JV9wAr7bLx #QuantumComputing #ArtificialIntelligence #MachineLearning #ML https://t.co/UNjYF6AK7r
Alibaba Unveils New AI Chip as Nvidia Access Remains Stalled
Alibaba unveiled its Zhenwu M890 AI accelerator, claiming three‑times the performance of the earlier Zhenwu 810E and positioning it as a domestic alternative while Nvidia’s H200 chip remains stalled in China due to U.S. export restrictions. The new chip joins a Zhenwu...
Data-Driven Decisions Accelerate AI Success
@rwang0 challenging our ideas about #ai by sharing many examples of how the pace of change, costs, & decision-making impact our success. do you have the data to make precise decisions? #CornerstoneConnect26 https://t.co/bKGFNihbHZ
Rethinking AI: Why Conventional AI Is Failing the Grid, and What a New Architecture Does Differently
Utilities face mounting pressure to inspect more assets with fewer specialists, and traditional computer‑vision AI—relying on massive labeled datasets and periodic retraining—fails to keep pace with rare, high‑severity fault types. eSmart Systems’ new white paper proposes Adaptive AI, a patent‑pending...
Gemini 3.5 Flash Update Makes Google AI More Useful for Developers
Google has launched Gemini 3.5 Flash, a generally available AI model optimized for coding and autonomous agents. The model supports multimodal inputs—text, image, video, audio and PDF—and can execute code, manage files, and run command‑line tasks via the Gemini API....
How to Build a Service Team That Thrives Alongside AI
AI agents have moved from experimental projects to core components of contact‑center operations, with 66% of service organizations already deploying them. By automating routine requests, these agents free representatives to tackle complex problems and build deeper customer relationships. Leaders who...
Pharma Roundup: Bristol Myers Squibb Enters Strategic Agreement with Anthropic, Incyte and Genesis Expand Strategic Collaboration
Bristol Myers Squibb (BMS) signed a strategic agreement with Anthropic to deploy its Claude agentic AI platform across the company's research, clinical, manufacturing and commercial functions, giving more than 30,000 employees access to the technology. The partnership will focus on...
ChatGPT, Claude, Gemini and Grok Are Not Ready to Brief American Voters
Recent studies confirm that leading AI chatbots—ChatGPT, Claude, Gemini and Grok—still fail to provide accurate news citations, misattributing sources in more than 60% of queries and delivering misinformation in roughly 35% of election‑related prompts. The findings, first reported by the...
AI‑Powered WhaleSpotter Network Launches in San Francisco Bay to Cut Ship Strikes
The WhaleSpotter AI system went live this week in San Francisco Bay, scanning up to two nautical miles for gray‑whale blows and heat signatures. Operators from the Bay Ferry and marine scientists say the real‑time alerts will let vessels reroute...
Omio Taps Appier’s Agentic AI to Acquire Users in 21 European Markets
Omio has teamed up with AI‑as‑a‑service firm Appier to roll out Agentic AI‑driven paid‑media campaigns across 21 European markets. Within twelve months the travel platform expanded from a single‑country test in Spain to a continent‑wide growth engine, consistently hitting cost‑per‑acquisition...
AI Consolidation Begins: Blackstone & Google Forge an AI Empire
Private‑equity giant Blackstone and Alphabet’s Google have launched a $5 billion equity‑financed joint venture to build a U.S. AI cloud platform, targeting 500 MW of capacity by 2027 and offering Google’s Tensor Processing Units as a compute‑as‑a‑service. The partnership creates a debt‑free,...