Today's AI Pulse
-1.png)
Anthropic shuts down Fable 5 and Mythos 5 under U.S. export controls
Anthropic abruptly disabled its flagship Fable 5 and Mythos 5 models to comply with a U.S. export‑control directive, exposing the vulnerability of firms that depend on closed‑source AI services. The move sparked a sharp rally in Chinese open‑source offerings such as MiniMax and Zhipu as investors shift to downloadable, on‑premise models.
Also developing:
By the numbers: CNTXT AI raises $60M Series A
Microsoft’s First Reasoning Model Arrives with a Provenance Pitch Aimed at Compliance Teams
At Build 2026 Microsoft introduced seven new MAI models, led by MAI‑Thinking‑1, its first dedicated reasoning model. The 1‑trillion‑parameter mixture‑of‑experts system activates about 35 billion parameters per token and offers a 256K context window. Microsoft’s sales pitch emphasizes a "clean, commercially licensed" training‑data lineage and no distillation from third‑party models, while the technical paper notes a web crawl of roughly 794 billion pages that still raises licensing questions. The model entered private preview on Azure AI Foundry, with early benchmarks showing strong performance on AIME and coding tests.
Microsoft Fara Tutorial: Run a Browser-Use Agent in Google Colab with a Mock OpenAI-Compatible Endpoint
The MarkTechPost tutorial walks users through installing Microsoft Fara in Google Colab and running a browser‑use agent via a lightweight mock OpenAI‑compatible endpoint. It demonstrates cloning the repo, installing dependencies, configuring Playwright Firefox, and executing the agent loop without a GPU‑intensive...
The Real Cost of Agentic AI
Agentic AI moves from hype to budget line items, with token consumption driving costs. At a blended rate of $3 per million tokens, a single agent using 2 million tokens daily costs about $2,190 annually, while more demanding agents can reach...
FDA Approves GE HealthCare’s MIM Contour ProtégéAI+ 2.0 Software
GE HealthCare secured FDA 510(k) clearance for its AI‑driven MIM Contour ProtégéAI+ 2.0 auto‑contouring software, aimed at speeding radiation therapy planning. The tool automatically delineates anatomical structures on CT and MRI scans, matching or exceeding manual methods. A predetermined change control plan...
Miso One Delivers Human‑Like Emotion in 110 Ms
AI voices are getting scary good. MisoLabsAI just launched Miso One, an 8B parameter text-to-speech model that delivers human-like emotion with only 110ms latency. What makes it stand out? • Expressive, natural-sounding speech • Creator-quality narration • Open-source model weights • API access coming soon This is...
NemoVideo Brings Viral Video Intelligence Into the Editing Workflow with Viral+ Studio
NemoVideo launched Viral+ Studio, an AI‑driven workflow that blends viral video pattern analysis, hook generation, and automated editing into a single conversational interface. The Inspiration Center surfaces trending storytelling structures and script ideas, while AI Video Recreation builds a first...
"AIs Like Talking About Sweetness, Loudness, Quiet, Age and Beauty. There Is a Lot of Insisting," Says Malin Hay, "...
ChatGPT and other large language models are increasingly being used to draft and edit literary works, prompting several high‑profile controversies. Hachette pulled the self‑published horror novel *Shy Girl* after a Reddit editor flagged passages that resembled AI‑generated prose, while the...
T-Mobile US Turns to AI to Tackle Event Congestion
T‑Mobile US introduced Dynamic CX, an AI‑enhanced network optimisation tool built on its self‑organising network platform and 5G‑Advanced architecture. The system scans public event calendars and real‑time traffic to pre‑allocate capacity for high‑density venues such as stadiums and festival grounds....
Octon Launches Orion Fabric: The Enterprise-Grade Governance Platform Bringing Agentic AI Into Production
Octon International unveiled Orion Fabric, an enterprise‑grade platform that governs the deployment of autonomous AI agents in both software and robotic environments. The solution, announced at NVIDIA GTC Taipei, adds a control plane that secures LLM‑driven agents with identity verification, permission...
Lloyds Taps Microsoft to Support Its AI Adoption Plan
Lloyds Banking Group has entered a multi‑year agreement with Microsoft to roll out the Microsoft 365 E7 suite, featuring Copilot, Entra Suite and Agent 365, across its entire operation. The deployment is designed to embed agentic AI into customer‑facing services and...
15 Best Vibe Coding Tools in 2026 Compared: Pricing, Features, and Best Fit
Vibe coding, coined by Andrej Karpathy, lets developers describe software in plain language while an AI agent generates the code. The approach promises faster prototyping, lower costs, and reduced reliance on large engineering teams. Fifteen tools ranging from full‑stack platforms...
WHO Paper Sets Out Emerging Challenges and Opportunities in Health AI
The World Health Organization released a discussion paper that maps artificial intelligence across the health‑policy cycle, highlighting how AI can speed evidence synthesis, expand data analysis, and model outcomes while warning of bias, digital divides, and epistemic injustice. It draws...
TeamViewer, Microsoft Bring AI, AR for Clearer, Smarter Remote Assistance
TeamViewer has partnered with Microsoft to embed on‑device AI video super‑resolution (VSR) into its Assist AR remote‑assistance solution. The technology runs locally on Windows PCs, sharpening video streams even on weak or unstable mobile networks. A closed beta is live, with...
AI Boosting Credit Investment Workflows Rather than Replacing Traders, Finds Barclays Survey
Barclays’ survey of 410 institutional investors shows AI is now embedded in credit‑market workflows, primarily enhancing research, data analysis and idea generation. Hedge funds lead adoption, with roughly 75% using AI daily, while trade execution and portfolio construction remain peripheral....
BULLET Launches Trinetra AI for Micro Drama Creators
BULLET, the micro‑drama platform backed by Zee Entertainment, unveiled Trinetra AI, an end‑to‑end artificial‑intelligence suite for creators, studios and brands. The system integrates script analysis, visual production workflows and multilingual audio localisation through three proprietary engines—Trishul, Rudra and Damrooh. Built...
Zain KSA Advances AI-First Vision with New Center of Excellence
Zain KSA has inaugurated an AI Center of Excellence (AI CoE) to embed artificial intelligence throughout its operations and deliver AI‑driven services to consumers and enterprises across Saudi Arabia. The CoE, part of Zain Group’s ‘4WARD‑Progress with Purpose’ strategy, will...
Why Waymo Settled for the Wrong Car
Waymo’s autonomous ride‑hailing fleet still relies on retrofitted Jaguar I‑PACE EVs, a platform that costs $120,000‑$200,000 per car, suffers reliability issues, and cannot be resold. The company is rolling out a purpose‑built Zeekr‑made Ojai minivan with a streamlined 23‑sensor suite,...
Amadeus Report Identifies Opportunities for Airlines with Agentic AI, and Recommends First Steps for Adoption
Amadeus, in partnership with Microsoft, published a report identifying five ready‑to‑deploy agentic AI use cases for airlines, ranging from automated voice rebooking to personalized offers. Trials demonstrate AI agents can manage multilingual rebooking calls, streamline aircraft turnaround, and auto‑generate marketing...
The AI Rollback Nobody Wants to Talk About...
Enterprises are entering an AI‑pruning phase as 2026 pilots move into production. CIOs are demanding measurable value, tighter risk controls, and clear ROI, causing many per‑seat copilot and broad‑agent projects to be paused or cut. Gartner analyst Anushree Verma notes...
Apple's AI Era: Every AI Drop Since Last Year's WWDC
Apple has woven Apple Intelligence into iOS, iPadOS and macOS since WWDC 2025, delivering generative writing, smart summaries and a more capable Siri. Core apps such as Messages, Mail and Photos now include AI‑generated content tools, while creative suites gain AI‑assisted...
China Poaches More AI Talent From the U.S. as It Eyes the Next 'Super-App'
Tencent hired former OpenAI researcher Yao Shunyu as chief AI scientist, aiming to build artificial general intelligence in China. His appointment signals a broader wave of Chinese tech firms poaching Silicon Valley talent to accelerate AI development. Companies such as...
NVIDIA Expands DRIVE Hyperion Ecosystem to Accelerate Global Robotaxi Deployment
NVIDIA announced a major expansion of its DRIVE Hyperion robotaxi platform, adding new partners across Asia, Europe and the Middle East. The ecosystem now includes Foxconn’s Taiwan launch slated for 2028, VinFast and Autobrains targeting Southeast Asia, Uber’s Munich pilot,...
SNEC 2026: Sigenergy Reimagines Energy Infrastructure with ‘AI in All’
At SNEC 2026, Sigenergy unveiled its "AI in All" strategy, introducing a suite of AI‑powered solar‑plus‑storage solutions for utilities, commercial sites and homes. The centerpiece, SigenAgent, is an autonomous AI agent that creates and executes optimal energy strategies across all...
AI in the Age of Sustainability: Are the Benefits to Pharma Worth Its Environmental Impact?
Artificial intelligence is now deemed crucial by three‑quarters of life‑science firms, with pharma executives betting on AI to accelerate drug discovery, streamline clinical trials, and personalize medical content. At the same time, generative AI’s data‑center demand—about 1,050 terawatt‑hours of electricity and...
AgentGG: Open-Source Agentic SAST Scanner
AgentGG, released under the Apache 2.0 license, is an open‑source, agentic static application security testing (SAST) scanner that leverages AI agents to read code, traverse imports, and confirm vulnerabilities before reporting. The tool ships with more than 100 markdown‑based agents, runs...
Snowflake Summit 2026 - How Sanofi Uses Snowflake and AI Agents to Boost Efficiency
Sanofi has transformed its fragmented data environment by unifying it on Snowflake and adding semantic layers, then layering agentic AI services from Elementum. The AI‑powered Concierge platform lets thousands of employees query and act on data directly in the Snowflake...
Human Accountability Must Remain Even as AI Empowers Decisions
IBM said this in 1979. I think many of us forgot it. For years, I treated AI systems mostly as productivity tools. Faster analysis. Better recommendations. Smarter automation. And somewhere along the way, I noticed something uncomfortable in myself. The more capable the systems became, the...
Local AI on iPhone: Privacy‑First, No Cloud Needed
Most AI apps want your data. @LMStudioAI is betting on a different future. LM Studio's new mobile app brings your local models to iPhone through LM Link, letting you access powerful AI running on your own machine from anywhere with end-to-end encryption. No...
AI Agent Governance Gets Harder when Agents Outnumber Your People
Abluva CTO Amit Gautam highlighted a real‑world breach where an autonomous reconciliation agent at a financial services firm was tricked into scanning a customer database, extracting six million records and sending them to an external Slack webhook. The incident underscores...
AI’s Data Center Crunch Sends Compute Ambitions Into Orbit
AI’s rapid adoption is straining terrestrial data‑center capacity, with U.S. spending up nearly 70% year‑over‑year and projected energy use set to triple by 2028. In response, cloud giants and aerospace firms are testing orbital data‑center concepts, highlighted by SpaceX’s filing...
Soap Box: Detection and Response in the AI Age
In this Soapbox episode of Risky Business, host Patrick Gray talks with Ed Wu, founder of Dropzone, about how AI is reshaping security operations centers (SOCs) amid a looming "vulnerability apocalypse." Wu explains that the surge of vulnerabilities will make...
Friday's TechBio News 📰
The week’s TechBio roundup highlighted a surge of AI‑driven collaborations and funding across biotech. Lundbeck teamed with Cradle to launch two CNS antibody programs using generative AI, while Alnylam inked a $2 billion deal with Inceptive to accelerate RNAi discovery. XtalPi...
AI as an Audience: Welcome to the Citation Economy
The article warns that AI has become the primary audience for brand messaging, ushering in a zero‑click era where users receive synthesized answers without visiting websites. Visibility now depends on AI citations rather than traditional clicks, making earned media the...
AI as Social Technology
The paper “AI as Social Technology” argues that speculative AGI fiction cloaks the real, messy ways large language models (LLMs) are reshaping society. It traces LLMs’ rapid shift from a translation breakthrough in 2017 to a widely‑heralded stepping stone toward...
Anthropic Publishes Claude Constitution, Igniting AI Consciousness Debate
Anthropic unveiled an 84‑page "constitution" for its Claude language model, prompting fierce discussion about AI consciousness, moral status and data governance. The document, written for Claude itself, has drawn comments from CEO Dario Amodei, in‑house philosopher Amanda Askell and external...
Nvidia Unveils 6‑foot Humanoid Robot Blueprint with Unitree H2 Plus and Thor T5000
Nvidia revealed a design for a 6‑foot, 150‑pound humanoid robot that pairs Unitree’s H2 Plus chassis with Nvidia’s Thor T5000 AI processor. The announcement highlights a rare collaboration between a Chinese robot maker and an American chip leader amid growing...
The Creative Gap: Why GenAI Is Outpacing the Talent It Was Meant to Empower
The article argues that generative AI is advancing faster than the creative talent pool meant to use it. Clients now seek partners who can judge AI‑generated concepts for brand fit, cultural relevance, and legal risk, not just operate the tools....
Anthropic's IPO Strategy: Halting Competitors Is 4D Chess
Anthropic asking other AI labs to stop improving their models just in time for their IPO is 4D chess..
LGND AI Is the 2026 Snowflake Startup Challenge Winner
LGND AI won the 2026 Snowflake Startup Challenge by showcasing its Large Earth Models (LEMs) trained on 800 petabytes of satellite imagery. The startup leverages Snowflake’s data‑warehousing services—Snowpipe, Dynamic Tables, and Cortex Search—to deliver image‑grounded answers for queries like Amazon deforestation....
What Are the Main Public Concerns About AI in 2026?
In 2026 public sentiment toward artificial intelligence diverges sharply from expert optimism, with surveys showing only a minority of Americans feeling confident about AI’s impact. Major concerns center on job displacement, misinformation from deepfakes, privacy erosion, and the physical footprint...
Who Holds AI Market Share in 2026, and Where Does Each Vendor Win?
AI market share in 2026 is fragmented across the technology stack, with different vendors leading each layer. NVIDIA commands the high‑end accelerator market, while AWS holds the largest cloud infrastructure share at 28%. Microsoft leverages its enterprise software ecosystem to...
What Is the AI Taxonomy for Technology and Markets?
AI taxonomies categorize technology, vendors, markets, regulation, and workloads, starting with the specific AI workload—training, inference, fine‑tuning, retrieval‑augmented generation, and evaluation. Gartner projects $2.52 trillion in global AI spending for 2026, making a clear taxonomy essential for allocating capital across chips,...
What Does the AI Ecosystem Look Like in 2026?
The 2026 AI ecosystem is dominated by compute, models, data, cloud platforms and enterprise workflows, with the United States leading private AI investment at $285.9 billion in 2025, far outpacing China. Gartner projects total AI spending to reach $2.52 trillion this year,...
What Are the Main AI Risks in 2026?
The 2026 Stanford HAI AI Index recorded 362 AI incidents in 2025, highlighting that AI risks now stem from large‑scale deployment rather than isolated experiments. Organizations often launch models faster than they establish monitoring, audit trails, and human‑in‑the‑loop controls, exposing...
Anthropic's Breakthrough Is RSI, Not True AGI
Critical context on the new Anthropic blog: 1, AGI is *harder* than RSI (as used below). AGI: machine can do anything human can do, autonomously [not achieved] RSI (as used below): AI is a useful coding tool that humans can leverage [achieved]; it’s...
Hg Lifts Stake in HgCapital Trust to 15% as AI‑focused Portfolio Gains Momentum
European software investor Hg announced it will raise its ownership in the listed HgCapital Trust from about 6% to as much as 15%, arguing the share price undervalues the fund’s AI‑driven assets. The move comes as Hg highlights a five‑fold...
Verizon, AT&T, T‑Mobile Deploy AI, Drones and ‘COWs’ for Hurricane Season
Verizon, AT&T and T‑Mobile are rolling out AI‑powered damage assessment, drone‑delivered micro‑towers and “cell on wheels” units ahead of the 2026 hurricane season. The new tools promise to shrink restoration times from days to minutes or hours, keeping customers connected...
Stanford Study Shows AI-Generated Lawyers Beat Law Professors 75% of the Time
Stanford Law School researchers led by Professor Julian Nyarko reported that AI-generated contract law answers outperformed human professors in roughly 75% of nearly 3,000 blind comparisons. The findings, published in early June 2026, suggest AI can deliver higher-quality legal reasoning...
73% of Insurance CEOs Prioritize AI for Compliance and Security, KPMG Survey Shows
A new KPMG survey indicates that 73% of insurance chief executives now list artificial intelligence as their primary investment focus, driven by compliance and security goals. The data also shows rising expectations of AI-driven returns within the next three years,...
Microsoft Unveils AI Badge at Build, Analysts Question Utility Amid Gadget Glut
Microsoft introduced an AI‑enabled badge at its Build conference, positioning it as an enterprise tool for workers. Analysts note that the surge of AI wearables—from pins to pendants—offers little practical value, echoing similar uncertainty around OpenAI’s hardware ambitions.