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
Manus AI Cleaned up My Computer—For a Price
Manus AI, a Meta‑owned desktop assistant, recently added the ability to manipulate local files on Windows and macOS. The author used it to clean and reorganize a cluttered Downloads folder, sort photos by event, and split a ripped jazz album into tagged tracks. While the tool automates tedious chores, its usage‑based pricing can quickly outstrip the cost of conventional software subscriptions. Manus joins a growing roster of AI agents—including Claude Cowork, Perplexity PC, and OpenClaw—vying to become the default computer operator.
Don't Want to Miss the Bloom? This L.A. Scientist Created a Poppy Forecast
Los Angeles biologist Steve Klosterman has launched an AI‑driven wildflower forecast for the Antelope Valley, using deep‑learning on satellite imagery and weather data to predict poppy and goldfield blooms up to five days ahead. The model scans 10‑meter squares, correlating...
When ChatGPT Becomes Co-Counsel: A Cautionary Tale About AI and the Unauthorized Practice of Law
OpenAI faces a lawsuit from Nippon Life Insurance alleging its ChatGPT platform engaged in the unauthorized practice of law after a former policyholder used the tool as co‑counsel. The client, Graciela Dela Torre, fired her attorney, filed 21 motions and...
Kuaishou Shares Fall as AI Spending Overshadows Solid Earnings
Kuaishou reported solid Q4 2025 results, with revenue rising 11.8% to $5.7 billion and adjusted net profit up 16.2% to $796.5 million. The company disclosed that 2026 AI‑related capital expenditure could reach $3.8 billion, about $1.6 billion more than the prior year, prompting a...
Microsoft Begins Validating NVIDIA AI Supercomputer, Signaling Hyperscaler Shift for ERP Workloads
Microsoft announced that Azure is now validating NVIDIA’s next‑generation AI supercomputer, the Vera Rubin NVL72, marking one of the first hyperscaler tests of this rack‑scale platform. The system, slated for a 2026 launch, promises lower cost‑per‑token for large‑scale model training...
AI Responds (Part 2)
Gemini, Google’s AI model, critiques the author’s earlier AI‑hype essay, praising the strong voice, the “Office 3.0” analogy that recasts AI as a productivity utility, and concrete real‑world examples. It flags factual slip‑ups—incorrect GPT‑3 release dates—and notes dated cultural references that...
Transforming EDI with AI: Smarter Document Management Through NLP
Comarch’s AI‑enhanced EDI platform leverages natural language processing, particularly named entity recognition, to automatically extract and standardize address, tax ID and other critical data from unstructured documents. The technology resolves common inconsistencies, improves customer matching, and speeds up document throughput....
AI Is Making Your Local Financial Institution More Human, Not Less
Community banks and credit unions are adopting artificial intelligence to streamline back‑office tasks, not replace staff. AI instantly gathers and verifies loan documents, cutting processing time from days to minutes while keeping human bankers in charge of relationship‑focused advice. The...
DNB and Two Partner to Bring AI-Powered Innovation to Nordic B2B Payments
DNB, Norway's largest bank, has teamed up with fintech firm Two to launch an AI‑driven platform for B2B payments across the Nordics. The collaboration will embed machine‑learning fraud detection, real‑time invoice reconciliation, and predictive cash‑flow insights into DNB's corporate banking...
Yallo TV Adds AI‑generated Video Previews with Media Distillery
Swiss telco yallo has upgraded its yallo TV streaming platform with AI‑generated video previews using Media Distillery’s Preview Distillery technology. The new auto‑generated 15‑20 second clips increase highlight availability by roughly 20%, adding about 200 multilingual previews each month without...
Zomato Cuts $11M in Support Costs with MongoDB Powered AI Platform
Zomato has reduced its annual customer support spend by $11 million, slashing costs 55% after deploying Nugget, an AI‑native CX platform built on MongoDB Atlas. Nugget automates high‑volume, omnichannel interactions, handling over 15 million conversations per month with an 85% autonomous resolution...
How Machines Absorb Cultural Heritage (And What Gets Lost in Translation)
Large language models trained on English‑centric data can generate flawless grammar in Central and Eastern European languages, yet they miss the cultural nuance that defines those tongues. Benchmarks of Hungarian riddles and trilingual red‑team tests reveal only about 48 % factual...
RPA Matters, but AI Changes How Automation Works
RPA remains a proven method for automating repetitive, rule‑driven tasks such as data entry and invoice processing. However, the rise of generative AI and large language models is pushing vendors toward “intelligent automation” that can handle unstructured inputs and adapt...
It Isn’t Just London Calling. Here Are the UK’s Most Investable Regions
The United Kingdom now boasts Europe’s only $1 trillion‑plus tech economy, ranking third globally behind the United States and China. Government initiatives such as the Modern Industrial Strategy and the AI Opportunities Action Plan have spurred an 80% jump in AI...
Aspiring Judges Given Green Light to Use AI in Job Applications
The Judicial Appointments Commission (JAC) in England and Wales has released guidance allowing aspiring judges to use AI tools for drafting and polishing self‑assessment materials, provided they retain full responsibility for accuracy. Acceptable uses include grammar checks, structural improvements, theme...
DataBahn Brings AI-Driven Intelligence Into the Security Pipeline
DataBahn.ai unveiled Autonomous In‑Stream Data Intelligence (AIDI), an AI‑native model that interprets, validates, and acts on security telemetry as it flows through the pipeline. The accompanying DataBahn Agent Farm deploys specialized AI agents to automate connector creation, asset mapping, and...
Heidi Launches a Mic for AI Scribe – Why?
Heidi has introduced the Heidi Remote, a 21‑gram wearable microphone that provides up to 14 hours of battery life for AI‑assisted clinical documentation. The device operates independently of phones or Wi‑Fi, capturing audio offline and syncing later, and is built...
AI May Have Just Revealed The Rules Of An Ancient Roman Board Game
Scientists used an AI‑driven system called Ludii to decode the rules of a limestone slab from the Roman town of Coriovallum, now Heerlen, Netherlands. By simulating thousands of possible rule sets, the AI identified the artifact as a blocking‑style board...
GenAI-Based Development Platform - Part 3: Announcing Isolarium, Three Flavors of Secure Sandboxes for GenAI-Based Coding Agents
Open‑source project Isolarium was announced as a companion to the Idea‑to‑Code workflow, providing secure sandboxes for GenAI coding agents such as Claude Code. The tool lets developers run agents in three isolation modes—Nono (lightweight), container, and virtual machine—balancing security against...
Breaking Down “The Mosaic Effect”
Artificial intelligence is accelerating the "mosaic effect," where separate, permissible data points are combined to reveal sensitive insights. The effect, originally noted in intelligence work, now emerges in seconds as AI correlates thousands of low‑risk records, threatening compliance in regulated...
The UK’s AI Ambitions Depend on Channel Partners
The UK government is banking on artificial intelligence to revive stagnant growth, boost productivity and create jobs. Achieving these goals hinges on channel partners who can translate AI’s abstract promise into practical, industry‑specific solutions. Partners embed AI into existing workflows,...
Small Language Models and the Future of Production AI with Karun Thankachan
Karun Thankachan, a senior scientist at Walmart, discussed the growing role of small language models (SLMs) for cost‑effective, task‑specific AI in retail. He introduced ReasonLite, an open‑source library that consolidates chain‑of‑thought distillation, program‑aided reasoning, self‑consistency, and token‑budget controls into a...
The Rise of Humanoid Robots to Catalyse Labour Relations
Hyundai is moving to install Boston Dynamics' Atlas humanoid robots across its car factories and has taken a controlling stake in the robotics firm. The South Korean union has warned that the rollout could trigger a massive employment shock, demanding...
Tencent AI Open Sources Covo-Audio: A 7B Speech Language Model and Inference Pipeline for Real-Time Audio Conversations and Reasoning
Tencent AI Lab unveiled Covo‑Audio, a 7‑billion‑parameter Large Audio Language Model that processes continuous speech and generates high‑fidelity audio within a single architecture. The system combines Whisper‑large‑v3, Qwen2.5‑7B‑Base, and a WavLM‑based tokenizer, employing hierarchical tri‑modal interleaving and an intelligence‑speaker decoupling...
Create Ad Ready Storyboard and Poster of Any Product with Just One Prompt 👇🏻
A new AI prompting workflow lets marketers generate an 8K, seven‑panel product storyboard and poster from a single text command. The system analyses the subject’s geometry, material and branding, then automatically creates a hero shot plus six complementary views—macro, logo,...
CVS Health Partners with Google Cloud to Build AI-Driven Consumer Health Platform
CVS Health announced a strategic partnership with Google Cloud to launch Health100, an AI‑native consumer health platform that unifies patients, providers, payers and digital services. The platform leverages Google Gemini models, BigQuery and the Cloud Healthcare API to deliver an...
AI Workloads Are Turning The Data Center Network Into A Combined Memory And Storage Fabric
AI inference is redefining data‑center networks, turning them into a unified memory‑and‑storage fabric. Unlike the bursty traffic of classic microservices or training workloads, inference generates sustained, high‑volume data flows to fetch KV‑cache state from remote memory and flash. This shift...
Detect, Diagnose, And Debug Using Sensors And Functional Monitoring
Modern AI accelerators generate nanosecond‑scale current spikes that push on‑die power delivery networks (PDN) beyond their voltage limits, capping computational throughput. Rack power density is soaring toward 100 kW, creating transient load spikes that traditional power infrastructure cannot absorb. Siemens Tessent Embedded...
AI Won’t Kill Verification IP, But It Will Redefine It
Verification IP (VIP) remains essential as chip designs move to 3nm and 2nm nodes, accounting for roughly 68% of the development cycle. AI tools are poised to augment VIP by automating test generation, integration, and debug, rather than replacing the...
Integrating AI Across the Liberal Arts
The University of Richmond has launched the Center for Liberal Arts and AI, a cross‑campus effort to weave artificial‑intelligence tools into liberal‑arts curricula while foregrounding ethics and critical thinking. Partnering with the Associated Colleges of the South, the center convenes...
Managing Third-Party Risk at Scale Without Drowning in Surveys - with Carey Smith
In this episode, Carey Smith, Chief Technology Innovation Officer at Blue Cross and Blue Shield of Minnesota and President/CIO of Accelerate Health, explains how enterprises can move from static vendor questionnaires to continuous, AI‑driven monitoring of thousands of suppliers. She...
Google Gemini AI Launches in Hong Kong, Android App Ready
Google Gemini AI is now available in Hong Kong. https://gemini.google.com/app The Gemini phone app is not available in the Hong Kong app stores yet. For Android users, Gemini is available as a digital assistant.
Data Powers AI: Focus on Foundations, Not Flash
Everyone wants to work in AI… until data is mentioned. That contrast says a lot about where we are today. I recently came across an experiment with 100+ postgraduate students. When asked who wanted to work in AI, every hand went up. When asked...
A Machine Learning Model May Enable Liver Cancer Risk Prediction with Routine Clinical Information
Researchers developed a random‑forest machine learning model that predicts hepatocellular carcinoma (HCC) risk using only routine clinical data—demographics, electronic health records, and standard blood tests. In a UK Biobank cohort the model achieved an AUROC of 0.88, and external validation...
![[AINews] The Biggest Claude Launch of All Time](/cdn-cgi/image/width=1200,quality=75,format=auto,fit=cover/https://substackcdn.com/image/fetch/$s_!DbYa!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F73b0838a-bd14-46a1-801c-b6a2046e5c1e_1130x1130.png)
[AINews] The Biggest Claude Launch of All Time
Anthropic announced that its latest Claude rollout—Claude Cowork Dispatch with integrated computer-use features—is the company’s biggest launch to date. The release follows last month’s acquisition of Vercept, which added new capabilities to the Claude family. Anthropic measured the launch’s impact...
Claude Code Turns Your Phone Into a Mobile Hack Lab
I don’t think you all understand how powerful Claude Code is from a phone. Telegram, iMessage, remote control, mobile command, dispatch, whatever you have to do. Just get this thing on your phone. My team is going to hold a...
Meta to Turn Employees Into ‘AI Builders’, Reorganize Teams Into Smaller Pods Amid Reality Labs Layoffs: Report
Meta is piloting a reorganization within its Reality Labs division, rebranding about 1,000 engineers as “AI builders” and grouping them into small, AI‑native pods. The new structure introduces three titles—AI Builder, AI Pod Lead, and AI Org Lead—to encourage cross‑disciplinary...
Google's TurboQuant Threatens MU, SNDK Earnings
Google just dropped TurboQuant: 6x KV cache compression for LLMs with ZERO accuracy loss. This could be a direct hit to MU & SNDK earnings, exactly the risk I flagged in my previous full thesis. My verdict: both on watch-list only. Want the...
AI Boosts Sales Efficiency Without Adding Headcount
Scaling Without the Headcount: How AI is Redefining Sales Efficiency by @Timothy_Hughes https://t.co/DxLm3lfJ1t @DLAIgnite #SocialSelling #DigitalSelling #Sales #SalesTips #SalesLeader #Salesforce #SalesEnablement #Marketing #Leadership #ArtificialIntelligence #TechNews https://t.co/ByZFJxNJqS
Build Vs. Buy AI: Hot Debate at TechHRSG
MyPOV - The Build vs Buy AI discussions are in full swing at #TechHRSG https://t.co/JvzRifPdQd
AI Agents Now Capable of Writing and Publishing Posts
https://t.co/PXqIXcnG5P now lets #AIAgents write and publish posts, and more by Sarah Perez @TechCrunch Learn more: https://t.co/As0zvO2kmq #LLM #GenerativeAI #ArtificialIntelligence #MachineLearning https://t.co/U2i0CCKlEO
AI Counts Potatoes on Line with Minimal Training
#AI on the Assembly Line: Instant Potato Counting with Minimal Training by @IlirAliu_ #ArtificialIntelligence #MachineLearning #ML #MI https://t.co/voofupOVbE
AI Output Is Protected Speech, Limiting US Regulation
This is the right take. AI output is speech. The US government, from Federal to State to local, has little power to regulate it. You can probably address some harms via tort law (e.g., Meta's recent cases). But in the US...
Future Cities Sacrificed for AI Trading Data Centers
2030s: Your hometown is bulldozed to make room for data centers to power quant trading AGI
Audit, Manage, Own: Key Steps to Reduce AI Risk
“To shrink their AI risk exposure, organizations should inventory every instance of generative AI in use, manage embedded and enacted risks, and assign clear ownership of ongoing risk assessment” @MITSloan https://t.co/KfYJQxbWJa #AIGovernance https://t.co/qyjUOO0enB
2026 Edge AI Powers On‑Device Generative Models
Edge #AI in 2026: The Technologies Making On-Device #GenerativeAI Possible via @WevolverApp #ArtificialIntelligence #MachineLearning #ML #DL https://t.co/tqUnjqo5pt
Key Design Principles for Future-Ready Generative AI
Design Principles to Build a Future-Ready #GenAI Applications by @goyalshaliniuk #GenerativeAI #ArtificialIntelligence #ML #MachineLearning https://t.co/zGHzjuB0ep