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
DHS Plans XR Smart‑Glasses Pilot for ICE Agents, Sparking Privacy Outcry
The Department of Homeland Security is preparing to field a custom extended‑reality smart‑glasses system for ICE agents by September 2027. The glasses will capture video, run real‑time biometric matches against federal databases, and overlay alerts, a plan that civil‑rights groups warn could turn field agents into mobile surveillance platforms.
Celonis and Deloitte Deploy AI‑Powered App to Cut SOX Compliance Costs by 10%
Celonis and Deloitte have introduced the Celonis SOX and Internal Controls Manager, an AI‑driven application that replaces manual Sarbanes‑Oxley testing with continuous data monitoring. A pilot at a global pharmaceutical firm automated more than 100 controls, lowered control costs by...
Kloudfuse 4.0 Launches AI‑Governed Observability and Workload Isolation for Enterprises
Kloudfuse announced the general availability of Kloudfuse 4.0, a platform that adds AI‑governed observability and independent scaling of ingestion, query and control planes. The upgrade targets enterprises facing tighter FIPS compliance, AI‑agent usage, and the need for cost‑effective telemetry at...
Microsoft Mulls Delay of 2030 Clean‑Energy Goal as AI Data Centers Spike Power Demand
Microsoft is weighing a postponement or abandonment of its 2030 “100/100/0” clean‑energy commitment after AI‑driven data‑center expansion has added roughly one gigawatt of capacity every three months. The move could reset expectations for big‑tech climate governance as power‑intensive AI workloads...
BlackRock’s South Korea ETF Records $409 Million Outflow as AI Stocks Propel KOSPI to 75% YTD Gain
BlackRock’s iShares MSCI South Korea ETF (EWY) saw a record $409 million net redemption, the largest in its history, as AI‑related stocks drove the KOSPI up 6% in a single session and 75% year‑to‑date. The $23 billion fund’s outflow highlights how sector‑specific...
Claude Code: Save 60-90% of Tokens
The post outlines practical techniques to slash Claude Code token usage by 60‑90 percent. It categorizes token‑saving strategies by common use‑case scenarios, allowing readers to target the most relevant optimizations. The author emphasizes that most consumption stems from prompt length,...
IMF Warns AI-Driven Cyber Attacks Could Trigger Funding Strains for Emerging-Market Banks
The International Monetary Fund warned that AI‑enhanced cyber attacks could cause extreme losses, strain funding and raise solvency concerns for banks and sovereign debt holders in emerging markets. The IMF’s blog highlighted the rapid diffusion of advanced models like Anthropic’s...
'Pharma, Not Really’: Top Young AI Talent Shuns Careers at Big Drugmakers
Top AI researchers like 33‑year‑old postdoctoral fellow Mazdak Abulnaga are increasingly turning away from legacy pharmaceutical companies. The article highlights a growing perception that big drugmakers are bureaucratic, data‑restricted, and slow to adopt cutting‑edge machine‑learning tools. In contrast, startups, academia,...
Goldman Sachs Quant Chief Says AI Could Make Markets Less Efficient
Goldman Sachs Asset Management’s quant chief Osman Ali says AI is reshaping investment research, enabling the firm to analyze roughly 15,000 stocks daily with machine‑learning models. While AI can improve price discovery in under‑covered small‑cap and emerging‑market equities, Ali warns...
Teradata Unveils Autonomous Knowledge Platform to Push Enterprise AI Agents Into Production
Teradata announced the Autonomous Knowledge Platform, a single system that blends production‑grade AI, analytics and data management. The platform lets enterprises run autonomous AI agents at scale across cloud, on‑premises and hybrid environments, aiming to shift AI projects from pilot...
Small Village Known for Its Globally-Renowned School Has One of South Africa’s Most Advanced Systems to Stop Car Theft
Hilton, a small KwaZulu‑Natal village famed for its elite boarding school, has deployed one of South Africa’s most advanced vehicle‑theft prevention systems. The community‑funded network uses SNIPR automatic number‑plate recognition cameras linked to OneSpace’s analytics and national stolen‑vehicle databases, covering...
271 Bugs Found in Firefox, Zero Written by a Human Attacker. What This Means for the Future of Safe Code...
Mozilla’s Mythos AI, built by Anthropic, scanned Firefox and uncovered 271 security‑sensitive bugs, all originating from machine‑generated code. The previous scan with a general model found only 22 issues, highlighting the power of purpose‑built AI for vulnerability research. The findings...
Temporal Launches Serverless Option for Durable Execution Platform
Temporal announced a serverless offering for its Durable Execution workflow platform at the Replay 2026 conference in San Francisco. The new model lets customers run fault‑tolerant workflows on a consumption‑based basis, targeting AI‑driven and long‑running applications.
GoLabs Deploys Unitree Quadruped Robots for 24/7 Security Patrols
GoLabs, a new robotics firm based in Plano, Texas, announced the launch of a commercial security line built on Unitree quadruped robots. The autonomous four‑legged machines offer 24/7 monitoring, elevator navigation and thermal imaging, positioning GoLabs at the forefront of...
AI Agents Poised to Automate End‑to‑end Scientific Research
AI is moving beyond assisting research to conducting parts of it autonomously. With agentic systems now able to plan, reason and use external tools, academic papers and scientific workflows can increasingly be automated end to end. The implication is profound. Science may...
Cloudflare Lays Off 1,100 Workers, Blames AI‑Driven Automation
Cloudflare announced it will eliminate 1,100 positions, roughly 20% of its workforce, attributing the cuts to a surge in AI‑powered tool usage. The layoffs come as the company reports a 34% revenue jump but a net loss of $22.9 million and...
The Agent Runtime Wars Have Begun. Is Your Website Ready? Via @Sejournal, @Slobodanmanic
In mid‑April Cloudflare and OpenAI each released new agent‑runtime platforms that let AI agents execute long‑running tasks with crash recovery, sandboxing, and persistent sessions. Cloudflare followed with AI Platform, AI Search, an Email Service beta, and database integrations, positioning its...
Anthropic’s Dreaming Feature Raises Trust Concerns for Enterprises
I am confused as to how anthropomorphic naming by Anthropic is helpful for corporate clients. Specifically, the new "dreaming" feature as noted in this Reuters article from a few days ago: ********** https://lnkd.in/eP6uDgga "Anthropic said on Wednesday that its "dreaming"...
Post‑AI Client Expectations: Why Predictive Insight Is the New Baseline
AI is reshaping accounting from a focus on precision and historical reporting to predictive insight. Clients now expect real‑time forecasts, anomaly alerts, and proactive guidance, pressuring firms to act as strategic partners. Firms that embed predictive tools and advisory touchpoints...
5% GPU Utilization: The $401 Billion AI Infrastructure Problem Enterprises Can't Keep Ignoring
Enterprises are poised to spend $401 billion on AI infrastructure in 2026, yet internal audits reveal average GPU utilization stuck at a dismal 5%. This under‑utilization turns costly GPU assets into depreciating liabilities, prompting CFOs to demand measurable returns. Recent VentureBeat...
Sage Touts New, Enhanced Products for Advisory, Payroll, More
Sage announced a suite of AI‑enhanced products aimed at accounting firms and their clients, including the expanded Sage Intacct Advisory with industry‑specific templates and automation. The company also launched Sage HCM, a unified HR and payroll platform, and introduced Sage...
Week in Review: CTV and AI Dominate Ad Tech Earnings, Major Distributors Back Microdramas, and Kraft Heinz Ramps Up Ad...
Ad‑tech leaders The Trade Desk, PubMatic and Magnite all highlighted CTV and AI in their Q1 earnings, with mixed market reactions. The Trade Desk’s shares dropped 15% despite 12% revenue growth, while PubMatic and Magnite posted modest revenue gains driven...
NewsCorp CEO: AI Licensing Beyond Transaction, Protect IP
More about AI and publishers -> NewsCorp CEO Robert Thomson touted its recent AI content licensing deal with Meta as greater than “purely transactional,” though he warned of other firms illegally scraping the company’s content “IP powers AI,” he said. “IP is an...
AI Adoption Is Moving Faster Than Security Controls, Increasing Risk
Hospitality operators are racing to embed AI agents while security controls lag, leading to massive data breaches. In March 2026, attackers stole 527 credentials from Spanish and Austrian platforms, exposing personal data of more than 5 million guests. The article highlights...
The AI Voice Agent Stack That Books Calls and Closes Deals — Without a Sales Team
AI voice agents now handle inbound leads by introducing themselves, qualifying prospects, and booking calendar slots in real time, all without human sales reps. The lean stack—Vapi or Bland for calling, GPT‑4o or Gemini for language, Calendly/Cal.com for scheduling, HubSpot...
Arc Raiders Studio Confirms AI Anti-Cheat Is Tracking Player Behavior
Embark Studios released an AI‑driven anti‑cheat system for its multiplayer title Arc Raiders. The solution combines kernel‑level telemetry with machine‑learning models to detect abnormal player behavior, while a human team reviews every ban appeal. It also trains models to distinguish...
GetDandy Launches Autonomous AI Workforce for Local Businesses, Delivering 24/7 Customer Engagement With No Setup Required
GetDandy unveiled an autonomous AI workforce that manages every customer interaction for local businesses across phone, SMS, Google Business Profile, web chat, email and social media, requiring no setup or ongoing training. The system automatically trains agents using each business’s...
New Claude Plugin Generates Viral LinkedIn Posts Instantly
Last week, I spent 7+ hours building a set of Claude Skills to help you write viral LinkedIn posts faster. The result? A Claude Cowork plugin anyone can install to write publish-ready LinkedIn posts in minutes. Now, I originally built this to share...
New AI Search Links, Core Update Winners And Losers – SEO Pulse via @Sejournal, @MattGSouthern
Google rolled out five tweaks that add inline links, end‑of‑answer citations, and public‑forum previews to AI Overviews and AI Mode, giving each reference more context. A March core update analysis shows aggregators such as YouTube, Reddit and travel OTAs losing...
How Hershey Is Using AI to Future-Proof Its Marketing Model
Chocolate giant Hershey has built an AI‑powered marketing measurement system after nearly a year of development. Trained on three years of sales and marketing data, the tool aggregates social, search and streaming signals to deliver monthly mix updates and prioritize...
Eightfold AI Announces Integration of Agentic Interviewing with Oracle Fusion Cloud Recruiting
Eightfold AI announced that its autonomous AI Interviewer is now integrated with Oracle Fusion Cloud Recruiting, part of Oracle’s HCM suite. The integration embeds Eightfold’s skills‑based interview engine—covering screening, functional, and coding interviews—directly into Oracle’s recruiting workflows. It enables enterprise...
NVIDIA and Corning Expand U.S. Optical Manufacturing
NVIDIA and Corning announced a multiyear partnership to expand U.S. optical connectivity manufacturing for AI data centers. Corning will boost domestic capacity tenfold and increase fiber output by more than 50%, adding three new plants in North Carolina and Texas...
China Prioritizes AI Deployment, US Chases AGI Myth
Why China is Winning the Race: US chasing AGI myth while China builds the AI future US betting AI future on a goal no one can define, while China prioritizes mass deployment over big bang discovery "The real danger for...
The Secret to the Perfect CRM Is Building It Yourself with AI
Softr’s AI Co‑Builder lets businesses craft fully functional, AI‑powered CRM systems using plain‑English prompts, eliminating the need for code or developer resources. The platform automatically generates backend tables, defines data relationships, and configures role‑based permissions, streamlining setup and maintenance. Built‑in...
EU AI Act Deal Would Delay High-Risk Rules to 2027, Ban Abusive AI Content
EU lawmakers reached a provisional Digital Omnibus on AI that pushes the high‑risk Annex III obligations to Dec 2 2027 and the product‑embedded Annex I rules to Aug 2 2028. The deal also adds Article 5, a categorical ban on AI systems that generate child sexual abuse...
Founder Fridays No. 193
The post argues that AI is now rewriting the rules of every industry, turning costly, manual processes into closed‑loop, outcome‑focused businesses. It warns that traditional org charts and layered management become bottlenecks as AI can directly access and act on...
Newforma Outlines AI Roadmap and Ecosystem Updates
Newforma unveiled an AI‑focused product roadmap at Newforma World 2026, introducing Vojo, an AI assistant that provides natural‑language search, BIM model analysis, and AI‑assisted submittal reviews. The rollout includes Smart Email Filing, which automatically tags emails to projects and suggests...
OpenAI Puts GPT-5-Level Reasoning Into Voice Model
OpenAI unveiled three real‑time voice models—GPT‑Realtime‑2, GPT‑Realtime‑Translate and GPT‑Realtime‑Whisper—each powered by GPT‑5‑level reasoning. The APIs blend streaming transcription, instant translation and contextual reasoning in a single end‑to‑end service. Pricing for simultaneous interpretation is set at $0.034 per minute, dramatically undercutting...
The Deepfake Era Has Arrived and PR Is the Front Line
AI‑generated deepfake videos have moved from theoretical risk to active weapon in the 2026 U.S. midterm elections, exemplified by the National Republican Senatorial Committee’s fabricated ad of Texas candidate James Talarico. Similar attacks have hit Europe, with the Bank of...
AI Outpaces Regulators; Calls for Stronger Oversight Grow
The AI industry may be moving faster than governments can respond. Paul Tudor Jones says the U.S. is already late on AI regulation, especially around deepfakes and verification standards. The mood is changing quickly. Even many inside the AI ecosystem now appear...
How to Build Vector Search From Scratch in Python
The article walks readers through building a full‑stack vector search engine in Python using only NumPy. It explains how to generate or simulate 8‑dimensional embeddings, normalize them, and retrieve the nearest items via cosine similarity expressed as a dot product....
Align Intent and Skill Levels for Effective AI Training
If you're investing in AI training for your team, read this before you book anything. We've trained developers at dozens of companies to become AI engineers. And the biggest problem is never the content. It's the setup. Here are the two biggest...
Biologists Leveraging AI Will Outpace Those Who Don't
1/ AI won't replace you. But a biologist using AI will. Especially in bioinformatics, where the questions never stop coming. https://t.co/VYCu5ukfCT
How Dangerous Is Anthropic’s Mythos AI? | Bruce Schneier
Anthropic announced Claude Mythos Preview, a generative‑AI model so adept at spotting software flaws that it will be limited to a handful of enterprise partners. The UK AI Security Institute and independent tests show OpenAI’s GPT‑5.5 and smaller, cheaper models...
Early Misaligned AI Deals: Crucial yet Overlooked Opportunity
The potential to make deals with early misaligned AI is among the most important*neglected courses of action I know of at the moment.
Doing Real Work Signals Claude‑generated Text
You should add the phrase “doing real work” to the list of signs that the text you’re reading was probably generated by Claude.
Google Cloud Next ’26: Rubrik Secures and Accelerates AI Agents on Google Cloud
Rubrik unveiled Rubrik Agent Cloud (RAC) for Google Cloud’s Gemini Enterprise Agent Platform, delivering a unified control layer that secures and accelerates AI agent deployments. Powered by Rubrik’s Semantic AI Governance Engine (SAGE), the solution offers real‑time intent‑based guardrails, an...
AI Adoption in Turkey Skews Toward Large Firms, Broadband Hubs
Research @CEP_LSE on Turkey (2021–24) documents significant disparities in AI adoption which is concentrated among large firms & in regions with high-speed broadband & proximity to data centers particularly for software-intensive & cloud-based applications https://t.co/2wLHq4GrCL
How a Robot (Sort of) Made Me Lunch
London startup Kaikaku AI is piloting a food‑assembly robot called Fusion at a poke‑bowl shop, backed by $1.8 million in funding. The machine automatically dispenses vegetables, sauces and raw salmon into a bowl, while a human adds rice and final seasoning....
Cactus Music Taps AI for ‘Artist Ops’ to Support Human Musicians
Cactus Music, a Chile‑based agency that has handled over 450 Latin American releases, has launched Artist Ops, an AI‑augmented platform designed to streamline music marketing for independent artists. The freemium service provides basic tools for free, while professional and enterprise...