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
Hammerspace Posts Nearly 14x YTD Bookings Over Full-Year 2025 as AI Shifts to Inference and Physical AI
Hammerspace announced that its AI‑ready data platform is driving a surge in demand as enterprises shift from model building to operational AI. Bookings in 2026 are already nearly 14 times higher than the full‑year 2025 level, fueled by a need for instant inference and physical‑AI workloads. The company highlights a seconds‑to‑first‑token advantage, >95% gross retention and an NPS of 71, underscoring strong customer loyalty. Strategic alliances with Hitachi Vantara, Supermicro and Oracle Cloud Infrastructure broaden its reach across hybrid, sovereign and neocloud environments.
Superintelligent AI Not Required to Improve Cancer Outcomes
Big tech companies claim superintelligent AI could one day cure cancer. But do we really need smarter AI to change cancer outcomes? https://spectrum.ieee.org/can-ai-cure-cancer-javorsky?share_id=9456123
ChatGPT’s Politeness Comes From Human‑ranked Preference Tuning
ChatGPT doesn't sound helpful by accident. Someone taught it to sound that way. After the model learns to follow instructions, labs show it thousands of answer pairs. Humans pick which response feels clearer, friendlier, or more useful. The model learns the preferred style. This...
Digest: OpenAI Lays Groundwork For ChatGPT Ads in EU; Blackstone, H&F, Goldman Back AI Venture; Musk Settles Twitter Lawsuit Paying...
OpenAI is preparing to launch ChatGPT advertising in the European Union by adding a consent‑management pixel and updating its privacy policy to permit sharing purchase data with advertisers. Blackstone, Hellman & Friedman and Goldman Sachs are backing a new Anthropic‑led...
5 Fun Projects Using Claude Code
Anthropic’s Claude Code, an agentic coding assistant, can read, edit, and run code across environments. A KDnuggets article outlines five progressive projects—from a basic web app to a custom Model Context Protocol server—to teach developers how to prototype, build, and deploy...
Your GPUs Aren’t Slow, They Just Have a Short Memory
Graid Technology is tackling the growing KV‑cache bottleneck that plagues long‑context, agentic AI models. When the cache exceeds GPU HBM, it overflows to slower storage, causing latency spikes of up to 18× and throughput drops of tenfold. Graid’s KV Cache Server,...
Aligned Ecosystems, Automation, and Fast AI Drive Partner Success
#IBMPartner What separates the partners who thrive in the AI era from those who fall behind? Second year at IBM Think and that question felt like the throughline of everything today. Kareem Yusuf , SVP of Ecosystem & Strategic Partners...
Do We Really Need Smarter AI to Cure Cancer?
The debate over AI’s role in curing cancer intensifies as Emilia Javorsky argues that over‑investment in AGI/ASI distracts from practical AI tools and data infrastructure needed for personalized oncology. She stresses cancer’s heterogeneity, urging focus on early detection, trial acceleration,...
Exclusive: XCaliber Health Scores $6.5M for Workflow Platform
XCaliber Health announced a $6.5 million seed round to scale its agentic AI platform that automates healthcare administrative tasks. The system ships with pre‑built agents, analytics models and integrates with major EHRs such as Epic, Cerner and athenahealth. CEO Prakash Khot...
How to Use AI to Grow Your Business on Social Media in 2026
In this episode, Gary Vaynerchuk and his guest discuss how AI is becoming as essential as electricity for small business owners, using the example of dog trainers who want to boost their social‑media presence. They outline practical AI tools—like Runway...
The Great Chip Divide: AI Chip War Pivots From Hype to Execution
Intel cleared a U.S. antitrust review for its investment in AI‑chip developer SambaNova, bolstering its generative‑AI roadmap. The chipmaker also hired former Qualcomm executive Alex Katouzian to run its client computing and physical AI group, signaling a push into edge‑AI...
AI-Led Job Cuts Don’t Always Mean Stronger ROI — Gartner
Gartner’s survey of 350 global leaders using AI agents reveals that 80% have cut staff, some by as much as 20%, yet those reductions show no link to higher return on AI investment. The firm stresses that ROI is driven...
OpenAI Accelerates AI Phone with Dual NPUs, Custom SoC
OpenAI's version of Jarvis -> Kuo: OpenAI appears to be fast-tracking its AI agent phone with two NPUs and a custom MediaTek Dimensity 9600 SoC, targeting mass production as early as H1 2027 Real-world visual sensing: "The ISP is the headline spec,...
What the Harvard ER Study Says About O1 Beating Doctors at Diagnosis, Why It Means Differential Diagnosis Just Stopped Being...
A Harvard‑led Science paper pitted OpenAI’s o1 reasoning model against board‑certified physicians on 76 Boston emergency‑department cases. At the triage stage, o1 achieved roughly 67% diagnostic accuracy versus 50‑55% for doctors, and both groups climbed above 80% once full workup...
Meta's AI Scans Bone Structure, Not Faces, to Flag Kids
What could go wrong? :) -> Meta is using AI on Facebook and Instagram to detect under-13 users by analyzing bone structure, height, and visual cues, but says it's “not facial recognition” "We’re also adding visual analysis as a...
Security Vendors Must Own AI, Not Serve It
GenAI is going to augment nearly every layer of the security stack. The interesting question for vendors is not whether the stack disappears. It is whether you become a feature of someone else's AI, or whether AI becomes a feature of...
As AI Fuels Rapid Growth in Document Fraud, Leading Fraud Detection Startup Finovox Launches in the UK
Finovox, a European leader in document fraud detection, announced its entry into the United Kingdom as AI‑driven fraud spikes across the continent. A recent study shows the UK ranks third in Europe for document fraud, with 11% of respondents admitting...
Microsoft Report Claims AI Agents Will Reshape Organisations and Redefine Knowledge Work
Microsoft’s latest Work Trend Index, based on a survey of 31,000 workers in 31 countries, predicts that AI agents will become embedded participants in everyday workflows, giving rise to a new “frontier firm” model. The report says most knowledge workers...
AMA Presses Congress for Guardrails on AI Mental Health Chatbots
The American Medical Association (AMA) has written to congressional AI and digital‑health caucuses urging stronger federal safeguards for artificial‑intelligence‑driven mental‑health chatbots. The AMA warns that the rapid deployment of these tools is outpacing existing patient‑protection frameworks, creating risks of misdiagnosis,...
AMA Presses Congress for Guardrails on AI Mental Health Chatbots
The American Medical Association is urging Congress to enact stronger safeguards for AI‑enabled mental‑health chatbots, warning that current oversight lags behind rapid adoption. The AMA highlights risks such as misinformation, emotional dependency, privacy breaches, and harmful responses in crisis situations....
Lili CTO Liran Zelkha on Building AI that Disappears
Fintech firms have spent the past two years layering chatbots and in‑app assistants onto their platforms, but Lili CTO Liran Zelkha argues the effort is misdirected. He believes small‑business owners, who are time‑pressed and uninterested in new interfaces, need AI...
Google Says A New Wave Of AI Users Is Transforming Search via @Sejournal, @Martinibuster
Google’s search leaders Martin Splitt and Nikola Todorovic say a new wave of users is leveraging AI‑driven search for increasingly complex, multimodal queries. By moving AI from the background to the front of the experience, average query length and detail...
AI Threatens Human Minds Before Building Humanlike Robots
Honestly more worried A.I. will turn people into robots long before it turns robots into people. https://www.bailliegifford.com/en/uk/individual-investors/insights/ic-article/2026-q1-ai-isn-t-coming-for-your-job-it-s-coming-for-your-mind-10061431/
AI Investment Now Outpaces Dot‑Com Era Share
AI hardware and software investment is a larger part of our economy now than it was during the dot-com peak. https://t.co/PctEUUESxq
AI Search Is Transforming Discovery and Media Economics
Publishers are losing click‑through traffic as AI‑driven search results deliver answers directly on the page. FT Strategies’ “Future of Discovery” report maps this shift into four strategic models—niche specialist, intelligence provider, voice‑led brand, and mass‑reach publisher—each balancing owned versus platform...
AI Hardware Spending Surges 31%, Software 15%
AI spending on hardware is running at an annualized clip of 31% the past five quarters, while software is 15%. Both are well above longer trends. https://t.co/jpzfvBjcKR
Predict 2026 Corporate AI Spending After
Had a great ideas dinner @jpmorgan last night. To repeat a poll: Anthropic’s revenue is $30 to $40 billion annualized. Total corporate spending on AI will end 2026 at what annualized rate?
Machinex Releases MIND at IFAT 2026
Machinex unveiled MIND, a system‑wide artificial intelligence platform, at IFAT Munich 2026. The platform extends AI from individual machines to an integrated sorting ecosystem, delivering real‑time decision‑making, system optimization and advanced reporting. MIND powers new equipment such as the AirPulse...
Gemini Lets AI Draft Your Slide Deck First
🎨 What if you could hand off the first draft of your slide deck to AI? ✨ That's exactly what Gemini helps you do. 🔗 Full guide with examples on the blog. https://t.co/ItAW6Egcmr
Adapt to AI: Protect Your Entry-Level Career
#AI is wiping out entry-level jobs. Here’s how to surf the wave and not get crushed by it https://t.co/t08VOZcbss via @FastCompany
How Modern Accounting Teams Are Using AI Native ERPs
Campfire, an AI‑native ERP startup, showcased its AI agents in a live webinar hosted by founder John Glasgow. The session demonstrated how autonomous agents can handle flux analysis, accruals, and month‑end close tasks that traditionally rely on spreadsheets and manual...
LLMs' Unreasonable Effectiveness Makes Them Universally Powerful
The unreasonable effectiveness of LLMs is what makes them so weird. The labs don’t need to decide what kind of AI to build, because better LLMs do better at most things. Finance? Pig disease identification? Restaurant suggestions? Coding? Yup. Most tech doesn’t...
GenAI Moves From Speed to Measurable, Specialized Outcomes
The speed of new capabilities is not the only thing moving fast in the GenAI space. Use cases are shifting from broad “efficiency” narratives toward more specialized, measurable outcomes. @TheFuturumGroup https://t.co/PQpDMUXRzp
AI Adoption Outpaces Safety Policies, Leaving Organizations Exposed to Cyber Risk
AI tools are now ubiquitous in enterprises, with 90% of digital‑trust professionals confirming employee use. Yet only 38% of organizations have a formal, comprehensive AI safety policy and a further 30% operate with limited guidance, leaving a quarter without any...
LLM System Design Interview #42 - The Global Memory Trap
In a mock DeepMind interview, candidates are asked why a 5× increase in raw teraFLOPs yields only a 1.2× boost in end‑to‑end throughput. The correct answer points to the memory wall: GPU compute has outpaced global memory bandwidth, leaving the...
OpenAI’s Trillion-Dollar AI Bet Is a Study in ‘Riskmaxxing’
OpenAI has committed over $1 trillion to data‑center construction and compute contracts, including a $300 billion deal with Oracle and $250 billion with Microsoft. Its annualized revenue is about $25 billion, creating a 40‑to‑1 gap between obligations and earnings. Early‑2026 reports show the company...
AtlanTec Conference to Explore Advancing Responsible, Human-Centred AI
University of Galway will host the 2026 AtlanTec Conference on May 14, drawing more than 400 delegates to discuss responsible, human‑centred AI. The event’s theme, “Intelligence Augmented – the Frontier of Human Judgement,” features speakers such as Mark Little, Dr...
Behind the AI in the Newsroom: The Washington Post’s Vineet Khosla
Washington Post CTO Vineet Khosla explains the paper’s "AI everywhere" strategy, which embeds artificial intelligence across news production and consumer products. The newsroom has rolled out AI‑generated personalized podcasts, now exceeding 100,000 episodes, and internal tools like Haystacker that let...
New AI-Powered Age Assurance Measures to Place Teens in Age-Appropriate Experiences
Meta announced a suite of AI‑powered age‑assurance tools designed to automatically place teens in age‑appropriate experiences such as Teen Accounts on Instagram, Facebook and Messenger. The rollout adds visual‑analysis AI that scans photos and videos for age cues, expands the...
AI and Alternative Dispute Resolution (Are We Ready for AI-DR?)
The column warns that generative AI tools are increasingly used in the early stages of civil disputes, from drafting complaints to preparing for mediations, but they lack the ethical reasoning of human lawyers. Recent studies reveal that large language models...
Billion-Dollar AI Rounds Push April To Third-Highest Startup Funding Month In A Year
Global venture capital reached $56 billion in April, doubling year‑over‑year and marking the third‑largest month in the past year. AI dominated the market with $37 billion in funding, 66% of total venture dollars, driven by Anthropic’s $15 billion and Jeff Bezos’s Project Prometheus...
AI as Human Augmentation, Not Replacement
Leading figures in the AI industry are beginning to say that AI is a tool to augment humans and give us superpowers, rather than a successor species that will replace us. Good. https://t.co/sEeANoZ9Qu
Europe Is Building a System without a Physical Foundation
Europe is pouring billions into AI, semiconductors and data centres, yet its hardware backbone—advanced printed‑circuit‑board (PCB) manufacturing—remains largely offshore. Vytautas Ilgūnas, chief commercial officer of TLT PCB, warns that without a domestic PCB supply chain, the region’s technological sovereignty is incomplete....
Google: AI Makes Human Experience More Important For Content via @Sejournal, @Martinibuster
In a recent Search Off The Record episode, Google Search engineers Martin Splitt and Nikola Todorovic warned that AI is making basic factual content less valuable. As large language models can instantly retrieve specs and data sheets, the competitive edge...
Custom AI Diagnoses Unprofitable Traders, Offers Next Steps
If you have been trading for 6+ months and are not profitable yet, I built a custom AI to daignose why and give you next steps. reply to this post and I'll dm you it. (follow so I can actually DM you)
The Sequence Knowledge #854: Return of the King: Unrolling the xLSTM Architecture
xLSTM reintroduces recurrent gating into modern sequence models, combining LSTM’s memory control with block‑wise attention for parallel execution. The architecture delivers up to 30 % lower memory usage and roughly double the training speed of comparable Transformers while matching perplexity on...
SMEs Fall Short on AI Cyber Security Training Despite Rising Risks
AI adoption is accelerating among UK small and medium‑size enterprises, but cyber threats are rising in parallel. A new Moneysupermarket survey shows only 10% of SMEs provide AI‑specific security training, despite 43% of all businesses reporting a breach in the...
OpenAI's Symphony Boosts Developer Output Sixfold by Automating AI Agent Attention
OpenAI launched Symphony, a workflow engine that gives each development ticket its own Codex agent, sparking a sixfold jump in merged pull requests within three weeks. The system uses Linear as a state machine to let agents self‑assign, reducing the...
Are Those Brake Lights or a House on Fire? Your Security Camera Can’t Tell.
Home‑security firms are adding AI‑driven descriptive alerts that label what cameras see, turning generic motion notices into detailed narratives. Wyze Labs launched the feature in early 2025, charging $19.99 a month to its 13 million users, while Ring and Blink also...
HYBRD Unveils AI Coach "Brain" To Adapt Training Plans in Real Time
HYBRD launched HYBRD Brain, an AI‑powered coaching layer that rewrites athletes' training programs on the fly based on schedule changes, performance data and injuries, while costing a fraction of traditional personal‑coach fees. The platform aims to replace static, one‑size‑fits‑all plans...