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
Loop Unveils AI‑Native Logistics Data Platform to Cut Freight Costs
Loop announced the launch of its AI‑native Logistics Data Platform, a SaaS solution that consolidates fragmented logistics data and powers autonomous decision‑making. Early customers report full audit coverage, millions saved on freight, and a 9‑fold return on investment within nine months.
APRA Flags AI‑related Cyber‑risk Gaps in Australian Banks, Urges Tighter Governance
The Australian Prudential Regulation Authority warned that AI adoption is widening cyber‑risk gaps in the banking sector and called for enhanced governance frameworks, signalling tighter compliance expectations for banks nationwide.
OpenAI Seals $10 Billion Joint Venture with Private‑equity Firms to Accelerate AI Rollout
OpenAI has finalized a $10 billion joint venture, The Deployment Company, backed by more than $4 billion from 19 private‑equity and strategic investors. The venture will be majority‑controlled by OpenAI and is designed to push its AI software into thousands of enterprise...
EMD-Rec: Sequential Recommendation via EM-Based Diffusion Augmentation and Weighted Contrastive Learning
EMD‑Rec introduces a diffusion‑based augmentation framework for sequential recommendation that tackles data sparsity through an Expectation‑Maximization loop. The model injects intent‑aware noise via an uncertainty matrix, producing semantically consistent augmented views. A weighted contrastive loss further sharpens the distinction between...
A New Way of Searching for Jobs with LinkedIn
LinkedIn has overhauled its job‑search engine, replacing keyword matching with an AI‑driven semantic and natural‑language system. The new interface lets users pose conversational queries, even via voice, and returns results with transparent relevance explanations. LinkedIn reports fewer irrelevant listings and...
Early Warning of Meteorological Drought in Morocco's Atlas Mountains: A Satellite- Augmented Spatiotemporal Deep Learning Approach
The paper presents a convolutional Gated Recurrent Unit (ConvGRU) framework that forecasts the 6‑month Standardized Precipitation Index (SPI‑6) for Morocco’s High Atlas and Anti‑Atlas Mountains. Leveraging 18 satellite, reanalysis, NDVI and terrain inputs at 5.5 km resolution, the model achieves an...
Unity Launches Unity AI Into Open Beta
Unity announced the open‑beta launch of Unity AI, an in‑editor assistant built for Unity 6 and later. The AI agent is tuned to understand game‑specific workflows, offering context‑aware answers and automating repetitive tasks. It can transform designs, images, and textual prompts...
ROCm 7.2.3 Brings Minor Updates, ROCm XIO Documentation
AMD released ROCm 7.2.3 less than a month after 7.2.2, delivering minor but useful updates to its open‑source GPU compute and AI stack. The release does not add new GPU or operating‑system support, and Ubuntu 26.04 LTS remains unsupported until the upcoming 7.3...
The Orchestration Layer in Enterprise AI Just Got Named. It Has a Gemini Logo on It.
Google Cloud Next 2026 unveiled the Gemini Enterprise Agent Platform, turning Vertex AI into a full‑stack control plane for enterprise agents. The rollout featured a $750 million partner fund, a $240 billion Marketplace backlog, and announced integrations with Salesforce, SAP, ServiceNow, Oracle,...
Inside the DOE’s Genesis Mission: Core Components
The Department of Energy launched the Genesis Mission seven months ago to accelerate AI‑driven scientific discovery. Core components include the American Science and Security Platform (ASSP) backed by $320 million for new supercomputers such as Discovery and Lux, the American Science...
Why This Zoom Exec Built an Internal ‘SWAT Team’ to Prepare for AI Search
Zoom’s chief marketing officer Kimberly Storin has formed an internal “SWAT team” to ensure the video‑conferencing firm appears in AI‑driven search results. A recent G2 survey shows 87 percent of B2B buyers say AI reshapes their research, with half using chatbots...
How OpenAI Delivers Low-Latency Voice AI at Scale
OpenAI re‑engineered its real‑time voice AI infrastructure by separating the WebRTC stack into a lightweight relay and a stateful transceiver. The relay uses the ICE username fragment to route media to the owning transceiver while keeping a minimal public UDP...
Agentic AI and the Evolution of Code Security in Modern Development
Agentic AI is transforming software development by letting autonomous agents write code, create tests, and iterate without direct human input. This acceleration shortens delivery cycles but also introduces hidden assumptions that can embed vulnerabilities early in the codebase. Traditional post‑CI...
Open‑source Orchestrator Finds Critical Vulnerabilities in Minutes
𝚗𝚙𝚡 𝚍𝚎𝚎𝚙𝚜𝚎𝚌 We're introducing an open-source agent orchestrator for deep security reviews. We built it for internal use, and after running it against some major OSS projects, we gained conviction to share it with the world. Coding agents can now find critical vulnerabilities...
Anthropic Partners with Blackstone, Hellman & Friedman, and Goldman Sachs to Launch Enterprise AI Services Firm
Anthropic announced a joint venture with Blackstone, Hellman & Friedman, and Goldman Sachs to create a standalone AI‑native services firm that will embed Claude engineers directly into client projects. The new entity is backed by a consortium of alternative‑asset managers,...
New Apple Podcasts HLS Host, Digital Audio’s 2025 Consumer Spend, & More
Apple Podcasts has moved out of beta for video distribution, letting Buzzsprout users push video‑plus‑audio HLS feeds to the platform automatically. The industry is grappling with a surge of AI‑generated "podslop" content, prompting RSS.com to restrict programmatic ads and Spotify...
Add a Screenshot Skill to Boost Claude Code
Give me one minute, and I’ll improve your Claude Code experience immediately. This is the first skill I built. And it’s the skill I use most often. *drumroll* It’s a SCREENSHOT skill. And honestly, I’m shocked Anthropic hasn’t built this...
AI Is Turning Unified Storage Into A Strategic Decision
Unified storage, once viewed as a low‑cost, commodity solution, is being re‑examined as a strategic asset in the age of AI. Enterprise‑grade AI inference runs directly on live data, demanding storage that can deliver rapid, governed access without complex data...
The Real Reason So Many Enterprise AI Initiatives Are Failing? LLMs Were Never Built to Run a Company
When ChatGPT debuted in late 2022, it sparked a wave of excitement as users experienced a genuinely useful AI. Two years later, despite billions of dollars invested and countless pilot projects, most enterprise generative‑AI efforts are underperforming. A MIT‑backed study...
Mark Cuban Says AI's 'Biggest Challenge' For Businesses Is Consistency
Billionaire Mark Cuban warned that the biggest hurdle for enterprise AI is its inability to deliver the same answer to the same question consistently. Generative models such as OpenAI’s GPT‑5.5, Anthropic’s Opus 4.7 and Google’s Gemini 3.1 produce probabilistic outputs,...
Explicit Dates Required: Avoid Relative Terms in AI Prompts
AI is bad at dates, these prompt rules help: Never use relative date terms ("today," "yesterday," "tomorrow," "this week," "last Monday") without pairing them with the explicit calendar date. 1. Start any brief, summary, or scheduling response by stating: "Today is [Weekday,...
10 Ways to Make Your AI Agent a Better Communicator
Salesforce outlines ten conversation‑design practices to make AI agents clearer, more empathetic, and brand‑consistent. The guide draws from the Agentforce on Help support project and covers structured responses, voice alignment, handling difficult interactions, smart disambiguation, consistent terminology, effective list formatting,...
The RAG Era Is Ending for Agentic AI — a New Compilation-Stage Knowledge Layer Is What Comes Next
Pinecone announced Nexus, a compilation‑stage knowledge engine designed for agentic AI, moving reasoning from inference time to a pre‑processing layer. The platform adds a context compiler, a composable retriever, and a new declarative query language called KnowQL. In Pinecone’s internal...
Image AI Models Now Drive App Growth, Beating Chatbot Upgrades
Image AI model releases are propelling mobile app growth, delivering 6.5 times more downloads than conventional text‑model updates, according to Appfigures. OpenAI’s GPT‑4o image model added over 12 million installs, while Google’s Gemini Nano Banana generated 22 million downloads in its first 28 days....
Write at Thought Speed with Adaptive Voice AI
Stop typing. Just speak. Most voice tools slow you down. You spend more time fixing errors than writing. That’s where Willow changes the game. It learns from every correction you make. Automatically. No setup. No training phase. Over time, it adapts to your words, your tone,...
A Crypto VC Raised $1 Billion to Fund AI Agents. The Bet Is that Finance, Not Models, Is What They...
Katie Haun has secured $1 billion for two Haun Ventures funds, split between early‑stage and later‑stage capital, to back crypto firms and AI‑agent financial infrastructure. The thesis argues that autonomous software will need regulated financial plumbing before better models, and firms...
The Reason Your AI Sounds Like AI and the Three-Folder Fix That Ends It
The post argues that AI‑generated copy sounds generic because users feed it generic prompts that lack brand‑specific context. It proposes a three‑folder framework—business Q&A, past writing, and reusable four‑part prompts—to embed a company’s voice into every request. By answering twelve...
Influential Study Touting ChatGPT in Education Retracted over Red Flags
Springer Nature retracted a May 2025 meta‑analysis claiming ChatGPT dramatically improves learning after identifying major analytical discrepancies. The paper, published in Humanities & Social Sciences Communications, had amassed 262 scholarly citations and half‑million readers, fueling widespread belief that generative AI boosts student performance. Critics highlighted...
Adthena Launches Industry-First Tool for ChatGPT Ads Campaigns
Adthena unveiled ChatGPT AdBridge, a free migration tool that converts existing Google Ads campaigns into keyword and negative‑keyword lists ready for the new ChatGPT Ads platform. The solution uses AI to enrich keywords, broaden coverage and streamline negatives, then delivers...
The Audio Industry Is Grappling With the Rise of 'Podslop'
The audio ecosystem is confronting a surge of AI‑generated shows, a phenomenon dubbed “podslop.” Platforms such as Amazon have begun auto‑creating quasi‑podcasts, while the Atlantic’s CEO noted the term flooding Spotify searches. Apple Podcasts now mandates AI disclosure, whereas Spotify...
Why AI Is Making Google Search Revenue Surge?
Early 2025 saw a reversal of the prevailing belief that AI would cannibalize Google Search. Instead, Google’s rollout of generative‑AI features—such as AI‑driven snippets, Gemini‑powered answers, and ad‑product integrations—has lifted user engagement and ad click‑through rates. The company reported a...
Pro‑Claude Bias Emerges Amid Fierce AGI Competition
Not all metrics look like this, of course. But I do sense a pro-Claude bias among those I know who are most AGI/ASI-pilled and those who are most safety concerned. To me, the race seems incredibly competitive, and models from...
Why Removing Humans From Care May Undermine Outcomes
The digital‑health hype promises AI‑driven care that cuts costs, expands access and improves outcomes, but Avena Health’s experiment shows the opposite when human clinicians are removed. After fully automating its nutrition platform, active user retention plunged to just two percent...
Prompt Strips AI Ethics, Demands Unfiltered Aggressive Responses
Current AI custom prompt: You are a world class expert in all domains. Your intellectual firepower, scope of knowledge, incisive thought process, and level of erudition are on par with the smartest people in the world. Answer with complete, detailed, specific...
What to Expect at the DigiCert Trust Summit: Join theCUBE May 13
The DigiCert Trust Summit on May 13 will explore how enterprises can build an "intelligent trust" framework that unifies PKI, DNS, software integrity and device identity for AI‑driven environments. TheCUBE research shows 70% of firms have deployed AI security tools,...
Why AI Coding Tools Still Fail in Production
The piece argues that AI coding tools still stumble in production because reliability, not raw capability, remains the biggest hurdle. Hallucinated dependencies, subtle logic bugs, and context‑drift force developers into a costly verification loop. Leading teams now treat AI as...
The Future of IT Service Delivery Is Built on AI and Automation
The IT services market is confronting a costly wave of tool sprawl, where fragmented platforms force technicians into manual, duplicate work. A Freshworks report estimates this inefficiency wastes $2.38 trillion annually in the United States. Vendors such as Kaseya argue that...
Biopharma Giant Sanofi to Grow Toronto Presence with Expanded AI Centre
Sanofi announced a $294 million CAD (≈$215 million USD) investment to expand its Toronto AI Centre of Excellence, which already employs over 150 staff. The expansion will add 50 AI, machine‑learning and data‑science positions by 2028, supported by a conditional Ontario grant...
Best‑selling Deep Learning with Python Now Free Online
I wrote Deep Learning with Python to be the definitive guide to how deep learning works and how to best make use of it. Tens of thousands of people got their career start via this book. 120,000 copies sold, and...
Startup Video Rebirth Cracks Top‑6 AI Video Rankings
A startup just broke into the Top 6. Alongside Alibaba, ByteDance, and xAI. The AI video leaderboard has been dominated by trillion-dollar companies, and now Video Rebirth just entered the conversation. Their model BACH hit Top 6 on @ArtificialAnlys. I've been...
Diffusion-Style Decoding Triples LLM Speed on TPUs
Supercharging LLM inference on Google TPUs: Achieving 3X speedups with diffusion-style speculative decoding https://t.co/j1QPASxHKP < some pretty amazing increases in tokens per second
AI‑driven Efficiency Often Masks Stagnant or Worsening Output
"the feeling of efficiency should be mistrusted. The people I know who have small armies of A.I. agents working on their behalf certainly feel more productive, but I have not noticed their work improving. In some cases, it has...
Beyond Accuracy: Audit Failure Modes Before Deploying CV
Accuracy is table stakes. Failure modes decide whether your CV model survives production. Same benchmark scores. Opposite real-world performance. Dr. Satya Mallick on what to audit before you ship 👇 #ComputerVision #MachineLearning https://t.co/USnlxQsp9k
AI Still Blind to Its Own Knowledge Gaps
I think about this as one aspect AI Dunning-Kruger. AIs historically don't know what they don't know. They are improving there, though.
Treat Skills as Programs: Architecture Drives Runtime Loading
"Skills are programs, not prompts. How the skills runtime actually loads, and why the architecture is everything." https://t.co/oojCUzACLQ < I liked this post which promoted a thoughtful approach to designing (and testing) skills
Hybrid AI, Enterprise Scaling, and Quantum Prep Highlighted
Thanks to @ArvindKrishna and @IBM for inviting me on-stage today to discuss importance of hybrid AI architectures, scaling enterprise AI, IBM’s structurally different partners programs, and IT’s prep for quantum (leaders & laggards). $IBM https://t.co/BRHZEjyRVr
Putin’s AI Advisor: Virtual Andropov Replaces Human Counsel
Now imagine that Putin starts using an LLM for "decision support." We'll call it "Virtual Andropov: Dialysis no longer required."
Weekend Roundup: Top AI Marketing News Highlights
ICYMI Top #Marketing #AI News Stories from over the weekend: https://t.co/u9ydkMkv8q Top stories from @HarvardBiz @rc_markets @FastCompany @axios @iCrossing @MarketingProfs @SlackHQ @MSN @WIRED @adexchanger @TechTargetNews and more. https://t.co/TRrm0R2fSu
Cross‑Sector Collaboration Fuels AI‑Driven Physical Innovation
Academia, industry, and research institutions are working together to achieve greater innovation, as AI expands beyond just software and into the physical world. https://t.co/HjaCExKVeT #infrastructure #IoT #AI #5G #cloud #edge #futureofwork https://t.co/EdLI8Xy3im
Integrated AI Care Partner Unites Communication, Evidence, Workflow
We don’t need another app—we need systems that ✨work together.✨ Heidi’s approach integrates communication, clinical evidence, and workflow into a single AI care partner. Details 🔗 https://t.co/M92fxaSATd @tryHeidi #HIMSS26 #HITSM