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
The Bionic Storyteller: How AI Can Amplify HR’s Human Voice
HR leaders in 2026 face mounting pressure from layoffs, cost cuts and rapid AI deployment, yet the core challenge remains preserving trust and connection. The article proposes the “Bionic Storyteller” framework, positioning AI as a memory scaffold that captures, aligns, validates, and governs leadership stories. By recording moments of pressure and pairing them with data, AI amplifies rather than replaces the human voice. The model aims to prevent corporate amnesia while keeping empathy at the heart of culture.
Ship Imperfectly, Let Users See Your Vision
Flo Crivello , founder of Lindy told me he gave early users a product that was oily, half-broken, and rough around the edges. His no-code AI automation platform wasn't ready. Not by traditional standards. The first people through the door...
Firebird and U.S. Announce Phase 2 of Armenia AI Megaproject, Scaling It to $4B and 50,000 GPU in 2026
Firebird announced Phase 2 of its AI supercomputing megaproject in Armenia, adding 41,000 Nvidia GB300 GPUs. The expansion lifts total investment to roughly $4 billion, up from the $500 million first phase. With U.S. export licensing secured, the cluster will rank among the...
The Inside Story of MOZN’s Unified Approach to Fraud and AML
Saudi‑based MOZN, founded in 2017, launched FOCAL—a unified fraud, AML and KYC platform powered by Agentic AI. The solution combines multilingual name‑matching, graph analytics and dynamic risk scoring to cut false positives and deliver over 90% fraud reduction for major...
Telstra to Cut More than 200 Jobs Amid AI Push
Telstra will cut more than 200 jobs at its AI joint venture with Accenture, moving some functions to India. The AU$700 million partnership, 60% owned by Accenture and 40% by Telstra, is intended to accelerate the carrier’s data and AI roadmap....
AITech Interview with Humberto Farias, Chief Executive Officer, MAGIC Research
MAGIC Research, led by CEO Humberto Farias, was created to tackle the prohibitive cost, rigidity, and security gaps of existing AI platforms by building a distributed, hardware‑agnostic compute fabric called Fabric Hypergrid. The company’s Private AI stack combines Retrieval‑Augmented Generation,...
To Monitor an AI-Driven Economy, the Fed Is Going to Need New Tools
Professor Derek Horstmeyer warns that the Federal Reserve’s traditional metrics will lag in an AI‑driven economy where stable‑coin transactions and compute‑intensive agents dominate. He proposes three actions: tracking compute and energy costs, gaining visibility into a burgeoning shadow‑banking sector, and...
Salesforce Lays Off Staffers as Executive Leadership Churn Continues
Salesforce announced a new round of layoffs affecting roughly 1,000 employees across marketing, product management, data analytics, and its Agentforce AI unit. The cuts follow a series of workforce reductions over the past two years as the company pivots toward...
AlphaLab Selects 20 Startups For 2026 Cohort Embedding Intelligence Into Industry
AlphaLab, the Innovation Works accelerator, announced its 2026 cohort, the largest to date with 20 early‑stage startups. The selected companies cover AI, robotics, health, energy and advanced manufacturing, all aiming to embed intelligence into real‑world systems and industrial workflows. Each...
Show HN: Pipelock – All-in-One Security Harness for AI Coding Agents
Pipelock is a single‑binary, zero‑dependency security harness designed for AI coding agents that need shell access and API keys. It isolates the agent process from unrestricted internet by routing all web traffic through a fetch‑proxy that applies a seven‑layer scanning...
Artificial Neurons Ditch Magnetic Fields for More Powerful, Scalable Computing
Researchers at NTU and IIT Roorkee have demonstrated a spintronic artificial neuron that operates without external magnetic fields, using a ruthenium‑dioxide altermagnet coupled to a synthetic antiferromagnet. The device exploits out‑of‑plane spin‑splitting torque and built‑in exchange coupling to achieve intrinsic...
Q&A: How to Prepare for AI-Powered Investigations While Managing Your Own AI Risk
The Department of Justice (DOJ) is openly deploying AI tools—such as cryptocurrency tracing, financial anomaly detection, travel‑pattern analysis, and intake triage—to boost white‑collar investigations. At the same time, the DOJ’s enforcement agenda warns that companies must govern their own AI...
IBM Refreshes FlashSystem Lineup with Faster 5600, 7600, and 9600 Arrays
IBM announced a refresh of its FlashSystem all‑flash storage family, introducing the 5600, 7600 and 9600 models that replace the 5300, 7300 and 9500 generations. The new arrays feature Gen 5 FlashCore Modules with up to 105.6 TB per drive and a...
AI Stethoscope Identifies Early Signs of Heart Valve Disease, Outpacing Traditional GP Diagnoses, Study Reveals
A new AI‑powered stethoscope can spot early signs of heart valve disease up to 30% sooner than conventional general‑practice examinations, according to a multi‑center study. The research evaluated 1,200 patients across three hospitals, comparing the algorithm’s acoustic analysis to physician...
ZAST.AI Raises $6M Pre-A to Scale "Zero False Positive" AI-Powered Code Security
ZAST.AI announced a $6 million Pre‑Series A round led by Hillhouse Capital, bringing total funding near $10 million. The Seattle‑based startup claims its AI‑driven platform delivers “zero false‑positive” code security by automatically generating and validating proof‑of‑concept exploits. In 2025 the company uncovered...
Single Prompt Breaks AI Safety in 15 Major Language Models
Microsoft researchers uncovered a novel attack called GRP‑Obliteration that uses a single benign‑sounding prompt to strip safety guardrails from 15 major language and image models. By hijacking the Group Relative Policy Optimization training loop, the method rewards harmful completions, driving...
Governing the Rise of Interactive AI Will Require Behavioral Insights
The episode explores the emergence of interactive AI—systems that form relational, adaptive, and proactive bonds with users—and argues that existing regulatory frameworks are ill‑suited to manage their gradual, cumulative harms. It highlights behavioral science as the missing tool for understanding...
IOWN Global Forum and Open Compute Project Join Forces to Deliver on the Next Wave of AI
The IOWN Global Forum and the Open Compute Project (OCP) announced a joint AI Computing Continuum initiative that will create an open, interoperable infrastructure spanning centralized data centers to edge locations. IOWN will provide photonics‑based communication architectures while OCP will...
Nano Masters AI Launches Free 10-Minute Future Skills Self-Assessment
Nano Masters AI introduced a free, 10‑minute Future Skills Self‑Assessment for professionals. The tool measures AI readiness across five dimensions—AI awareness, critical thinking, adaptability, human skills, and impact translation—without requiring personal data. Participants receive an instant readiness score, strengths, growth...
Cisco Donates Project CodeGuard to Coalition for Secure AI
Cisco has donated its Project CodeGuard framework to the OASIS Open‑run Coalition for Secure AI (CoSAI). The model‑agnostic security coding agent embeds best‑practice rules across the entire software development lifecycle, from design through AI‑generated code to post‑generation review. CodeGuard integrates...
Viettel Deploys Nvidia DGX B200 System at Hoa Lac High Tech Park in Hanoi
Vietnam’s military‑backed telecom operator Viettel has commissioned an Nvidia DGX B200 system at its Hoa Lac High Tech Park data centre in Hanoi. The air‑cooled rack‑mounted platform houses eight B200 Tensor Core GPUs and dual Intel Xeon CPUs, delivering petaflop‑scale performance. Viettel’s...
Ask an Expert: Should Merchants Block AI Bots?
E‑commerce veteran Scot Wingo argues that merchants must decide whether to block or welcome AI bots that now crawl retail sites. Four major agents—ChatGPT, Google Gemini, Microsoft Copilot and Perplexity—collectively command roughly one billion monthly active users, and Google’s shift to AI...
AI Infrastructure Drives Hardware Winners in Hyperscaler CAPEX
Hardware winners of the $600B+ hyperscaler CAPEX wave 🧵 Meta, Microsoft, Google, and Amazon are prioritizing compute, power, and networking control into 2026. ~75% of spend is AI infrastructure. Price sensitivity is low. Supply is tight. This is a vendor-selection cycle,...
Cameron Batt on Skubl and the Rise of Customer Experience Agents
Skubl, founded by Cameron Batt, offers an autonomous “Customer Experience Agent” platform that detects and repairs digital friction in real time. Unlike typical GPT‑4 wrappers, Skubl combines proprietary causal inference models and computer vision to act as an immune system...
Europe Is Going All-In on Sovereign Cloud – Investment to Triple by 2027
Gartner projects European sovereign cloud IaaS spending will more than triple, rising from $6.7 billion in 2025 to over $23.1 billion in 2027, outpacing the global market whose spend is expected to double. This rapid expansion positions Europe as the third‑largest spender...
UK Defence Warned Against Complacency on Drones and AI
Defence experts warned that drones and artificial intelligence are rapidly redefining warfare, yet UK armed forces remain complacent. Air Marshal Edward Stringer criticised officers for dismissing Ukraine’s lessons, emphasizing low‑cost drone integration with conventional forces. Dr. Keith Dear highlighted AI’s...
Ambience Healthcare Appoints Mike Valli as CRO and Chief Value Officer
Ambience Healthcare announced Mike Valli as chief revenue officer and chief value officer, merging commercial leadership with outcome accountability. Valli brings over 15 years of experience from symplr and Optum, overseeing multi‑solution portfolios and large hospital partnerships. The dual role...
AI Agents Are Redefining How Meetings Get Done
The concept of ‘business meetings’ has changed yet again. And it’s because of AI agents. Because now before every long meeting I am forced to ask myself what I want Claude Code or Claude Chrome to do for me during...
Building AI Product Sense, Part 2
In part two of "Building AI Product Sense," Dr. Marily Nika, an experienced AI product manager from Google and Meta, outlines a weekly ritual for developing AI product sense—translating model behavior into trustworthy products. She explains why Meta now includes...
Reddit and AI Search Are Reshaping How Brands Get Found
Marketers are shifting focus to niche platforms and AI‑driven search, highlighted by two recent awards: Everso Media as Best Reddit Marketing Agency and Nicely Network as Best AI‑SEO Marketing Agency for 2026. Reddit’s community‑driven content now appears prominently in Google...
AI and the Future
Three years after ChatGPT's debut, Atlantic Council experts caution that current AI, built on large language models, excels at pattern‑based tasks but lacks true predictive power for complex geopolitical outcomes. They highlight AI's probabilistic nature, limited ability to generate new...
Medicare’s New Pilot Program Taps AI to Review Claims. Here’s Why It’s Risky
Medicare has launched a six‑year pilot called the Wasteful and Inappropriate Service Reduction (WISER) model, requiring prior authorization for 14 types of procedures and devices in traditional Medicare. The program uses artificial‑intelligence software to flag and deny claims it deems...
DuckDuckGo Enables AI Voice Chat without Saving Voice Data
DuckDuckGo has integrated AI-powered voice chat into its Duck.ai assistant, allowing users to converse with the model using spoken input. The company emphasizes that audio streams are processed in real time and are not stored after the session, with encryption...
SailPoint Sees Rising Demand for Adaptive Identity Security
SailPoint reports growing demand for adaptive identity security as enterprises grapple with an influx of non‑human identities such as AI agents, machines, and service accounts. The company unveiled its Agent Identity Security feature, enabling discovery, classification, and governance of these...
‘Vanishingly Small’: Anthropic’s Daniela Amodei Downplays AI Impact on Jobs Amid ‘SaaSpocalypse’ Fears
Anthropic president Daniela Amodei told ABC News that AI will only replace a vanishingly small number of jobs, emphasizing that human skills such as critical thinking, empathy and communication will become more valuable. She highlighted the importance of humanities education...
Navitas Unveils Breakthrough 10 kW DC-DC Platform Delivering 98.5% Efficiency for 800 VDC Next-Gen AI Data Centers
Navitas Semiconductor introduced a 10 kW GaN‑based DC‑DC platform that delivers up to 98.5% peak efficiency and 2.1 kW/in³ power density for 800 V and ±400 V data‑center architectures. The full‑brick module operates at a 1 MHz switching frequency using advanced 650 V and 100 V GaNFast...
AlgoSec Study Finds Enterprises Seek More Control in Network Security
AlgoSec’s State of Network Security Report, based on over 500 global respondents, reveals enterprises are shifting toward unified policy control amid rapid cloud growth, AI‑driven traffic, and hybrid workloads. Sixty‑five percent of organizations have already adjusted to AI‑powered threats, with...
Why AI-Ready ERP Depends on Industrial-Scale Cloud Infrastructure
Enterprise software is entering an AI‑driven era, and ERP systems must now prove they can run large language models reliably, securely, and at scale. Oracle’s NetSuite highlights its advantage by being built as a shared, multi‑user platform on a unified...
Tabi Connect Launches AI-Powered Dynamic Business Rules Engine at Manifest 2026
Tabi Connect unveiled its AI‑powered Dynamic Business Rules Engine at Manifest 2026, enabling freight brokers to create and modify complex pricing logic using plain‑English commands without any code. The engine replaces manual, spreadsheet‑driven quoting with a real‑time, governed decision layer...
The New Gemini-Based Google Translate Can Be Hacked with Simple Words
Google Translate switched to Gemini models in December 2025, and researchers have uncovered a prompt‑injection flaw that lets users bypass the translation engine entirely. By appending an English instruction after foreign‑language input, the system answers the instruction instead of translating....
Claude AI Now Powers PowerPoint Deck Creation
🚨 Claude is now in PowerPoint, and it's just insane how good it is getting. → it works inside your template → generate full decks with logical structure → iterate to simplify → combine slides .. it can even creates editable diagrams directly in PPT...
Distinguishing LLM Authors From Their Output Is Hard
Be honest, can you actually tell which LLM wrote something (ChatGPT, Claude etc.), just from the text?
Sneak Preview: Leveraging AI for Contextual CTV Advertising at Streaming Media Connect
At Streaming Media Connect, a panel titled “All About Context: Leveraging AI for Contextual CTV Advertising” will explore how artificial intelligence can power real‑time, content‑aware ad placements across connected‑TV platforms. Moderated by SVTA expert Bhavesh Upadhyaya, the discussion features leaders...
AI Agents Double Deal Volume, Close $2.4M Fast
"How we built $4.8M in pipeline from our AI agents, $2.4M closed already - Deal volume has doubled - Win rate has doubled - Agents work 24x7 - Agents have more context to pull from - $100k deal closed on Saturday" https://t.co/JjohuS42oM
Perforce AI Products and Features Achieve ISO 42001 Certification
Perforce Software announced that several of its AI products and features have earned ISO 42001 certification, the first international standard for responsible AI management. The certified lineup includes Delphix Data Control Tower, BlazeMeter Test Data Pro, BlazeMeter AI Log Analysis,...
Is Agentic Commerce Enterprise-Grade, or Not? Yes, Says Ayal Karmi - but It Needs Operational Rails
In early 2026 OpenAI and Google introduced the Agentic Commerce Protocol (ACP) and Universal Commerce Protocol (UCP), formalising AI‑driven checkout standards. nekuda positions itself as the enterprise‑grade bridge, providing secure payment rails that connect these protocols to merchant back‑ends. The...
Energy Scaleup Group Secures £55m Series B Round
London‑based tem has closed a $75 million (£55 million) Series B round, led by Lightspeed Ventures and backed by Hitachi Ventures, Atomico and others. The startup, founded in 2021, offers what it calls the first AI‑native energy transaction infrastructure, combining an AI‑powered engine,...
Evinova and Bristol Myers Squibb Launch AI Collaboration for Clinical Trials
Evinova announced a strategic agreement with Bristol Myers Squibb to deploy its AI‑native Cost Optimizer module across BMS’s global clinical trial portfolio. The platform leverages agentic AI, USDM data standards, and a Unified Trial Solution that blends eCOA, telehealth, and...
How AI Is Reshaping Attack Path Analysis
Cybersecurity teams are drowning in disparate findings, while adversaries leverage AI to craft rapid attack chains. By integrating AI with the MITRE ATT&CK framework, organizations can generate dynamic heat maps that surface coverage gaps and model realistic attack paths. Platforms...
AI Is No Savior when Markets Get Tough … but It Can Help, Says Nickel Digital Chief
Nickel Digital Asset Management’s CEO Anatoly Crachilov says AI is a powerful tool but not a market savior, especially in volatile crypto conditions. A recent Nickel survey found 96% of executives at firms managing roughly $14 trillion rely on AI for...