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
Are You Using A.I. at Work? We Want to Hear From You.
The New York Times is soliciting input from white‑collar workers about how they use artificial intelligence in their daily tasks. Reporters note that AI is now handling legal drafting, research synthesis, and marketing content across non‑tech sectors. The outreach aims to capture real‑world experiences, tool preferences, and challenges such as privacy and bias. Collected insights will feed future reporting, policy discussions, and corporate AI strategies.
Your LLM Is Ignoring Its Own Mistakes (And Three Papers That Show How to Fix It)
LLMs excel at generating first‑pass outputs but struggle to learn from real‑time feedback. Recent research—Meta’s RLEF, Anthropic’s Constitutional AI, and the ReAct framework—demonstrates that reinforcement learning, self‑generated critique, and explicit reasoning traces dramatically improve error correction and tool use. Across...
Why Reputation Is One of Your Most Valuable Assets
In a March 16 podcast, D S Simon Media CEO Doug Simon and 84 Communications founder Rina Foster examined how AI is reshaping reputation risk. They warned that AI‑generated images, audio and video make false information spread faster across platforms. AI‑driven monitoring...
Tesla's 'Fleet' Of Unsupervised Robotaxis In Texas Appears To Have Been Cut From 8 Cars To Only 4
Tesla’s unsupervised robotaxi fleet in Austin has shrunk from eight vehicles to roughly four, according to the Robotaxi Tracker. The reduction follows a series of safety incidents, including repeated failures to stop at active railroad crossings and a crash rate...
University of Calgary Researchers Use AI to Hunt New Treatments for Cattle Parasites
University of Calgary researchers have secured a $1.4 million NSERC grant to apply AI‑driven genomics in the hunt for new anti‑parasitic drugs targeting roundworms in beef cattle. By sequencing helminth genomes and screening large chemical libraries, the team has already identified...
The Supreme Court Declines to Answer AI’s Authorship Question—For Now
The U.S. Supreme Court denied certiorari in Thaler v. Perlmutter, leaving the D.C. Circuit’s ruling that works generated entirely by artificial intelligence lack copyright protection. This decision upholds the long‑standing requirement for human authorship under the Copyright Act. While the...
Time for Your “AI or Human?” Radio Announcement Quiz
In this brief episode, the host promotes The Rock Dog and its sponsor, Bear Carpet One, highlighting the company's flooring options and its range of window treatments, including Hunter Douglas plantation shutters and honeycomb blinds with power opening and Bluetooth...
Fiscal AI Partners with Aiera to Power AI‑driven Research
I'm pleased to announce @fiscal_ai is partnering with @Aiera to continue to build the future of investment research in the AI era. Aiera empowers research providers to embrace AI while retaining full control over their intellectual property, while giving buy-side firms...
Unlocking the Power of Conversation: How Slack’s New Platform Is Fueling the Agentic Era
Slack unveiled a new platform layer featuring a Real‑Time Search API and a Model Context Protocol, giving developers secure, real‑time access to conversational data for AI agents. The tools let large language models pull context‑aware information directly from Slack channels,...
AMC Robotics and HIVE Announce Collaboration to Advance AI-Driven Robotics Compute Infrastructure
AMC Robotics (NASDAQ: AMCI) announced a strategic collaboration with HIVE Digital Technologies (TSXV: HIVE) to use HIVE’s GPU‑accelerated AI compute infrastructure for its expanding robotics development. The partnership already powers the Kyro™ quadruped robot, demonstrated at the Tokyo Security Show...
Threshold to Zero: Preserve High Pixels, Reveal Soft Edges
Understanding Threshold to Zero in Image Processing In Threshold to Zero, pixel values are kept only if they are above a chosen threshold - otherwise they are set to 0. The inverted version does the opposite: values above the threshold become...
What IT Leaders Can Learn From a Housing Authority’s AI Transformation
NWN relaunched its Intelligent Cloud service, a multi‑vendor control plane that adds observability and governance across hybrid environments. The platform enabled the South Mississippi Housing Authority to replace its legacy phone system with an AI‑driven Amazon Connect solution, deflecting 68 %...
US Deploys AI for Warfare, Iran Becomes Test Ground
The US turned to Claude and other artificial intelligence systems to help shape its vision for AI warfare. In Iran that strategy is unfolding in real time An excerpt from the coming book Project Maven shows how the US enlisted...
9 AI Prompts to Break Your AI Out of Default Mode
Most users rely on generic prompts that coax AI into its statistical default, delivering safe but unremarkable answers. The post presents nine specialized prompts that force AI out of this default zone by drawing on niche data, multilingual sources, and...
Advisor AI Adoption Rises to Over Half, Notetaking Gains
This #WeekendReading kicks off with the news that the latest @t3techhub/Inside Information Software Survey shows an uptick in advisor use of Artificial Intelligence (AI) tools, with approximately 52% of survey respondents using AI search and generative language functionality (up from...
Financial Statements Assume Human Customers. What Happens When AI Agents Drive Your Revenue?
AI agents are emerging as direct revenue sources for SaaS firms, yet most companies do not disclose how much income they generate. Because agent‑driven revenue can swing dramatically after a foundational model is retrained, traditional metrics like CAC and churn...
Jeonbuk National University Researchers Develop DDINet for Drug-Drug Interaction Prediction
Researchers at Jeonbuk National University have unveiled DDINet, a lightweight neural network designed to predict drug‑drug interactions (DDIs) for previously unseen compounds. The model employs five fully‑connected layers and molecular fingerprints, with Morgan fingerprints delivering the best results. Using a...
Preparing Enterprise Data Centers for AI Adoption
A 2025 McKinsey forecast puts AI‑related IT infrastructure spending at nearly $7 trillion through 2030, with $3 trillion earmarked for data‑center capacity. Enterprises must juggle this surge with traditional workloads, requiring distinct strategies for power‑intensive AI training and latency‑critical AI inference. Training...
BuzzFeed Nearing Bankruptcy After Disastrous Turn Toward AI
BuzzFeed’s 2023 pivot to artificial intelligence was heralded by a soaring stock price that jumped from roughly $3 to over $15 per share, as the company promised AI‑generated quizzes and articles. The experiment quickly faltered; AI‑driven content proved low‑quality, prompting...
These Stocks Will Benefit From AI and an Aging Population, Jefferies Says. They Also Pay Dividends
Jefferies analysts say artificial intelligence and an aging population are creating a powerful tailwind for senior‑housing REITs. They highlight Welltower, which has built a decade‑long data‑science platform, integrated OpenAI, and offers a 1.4% dividend, and American Healthcare REIT, a 2024...
Opus 4.6 1M Becomes Default for Claude Code Plans
🎁 Opus 4.6 1M is now the default Opus model for Claude Code users on Max, Team, and Enterprise plans. Pro and Sonnet users can opt in with /extra-usage.
Do Chatbots Fill You With Rage? This Startup Will Pay You $100 an Hour to ‘Bully’ AI.
Memvid, an AI memory startup, is paying $800 for an eight‑hour shift to "bully" popular chatbots, offering $100 per hour for a "professional AI bully" role. The job requires no computer‑science background; candidates simply need patience and a history of...
🔮 The Lantern and the Flame
In this episode, the host explores how pervasive AI tools are reshaping our thinking, distinguishing between helpful cognitive offloading and the dangerous trend of cognitive surrender—uncritical abdication of reasoning. Drawing on research, personal workflow experiments, and references to thinkers like...
Razer Flexes AI Desktop Companion and Other New Products at GDC
At the Game Developers Conference 2026, Razer showcased a suite of AI‑powered products, headlined by an AI desktop companion designed to streamline gaming workflows. The company also introduced AI‑enhanced peripherals, including a smart mouse and headset, that promise real‑time performance...
U.S. Tech Giants Flocked to the Persian Gulf. Now They Are Targets.
U.S. tech giants including Amazon, Google, Microsoft and OpenAI have poured billions into data center projects across the Persian Gulf to support AI development. Iranian drone attacks in early March damaged Amazon’s facilities in Bahrain and the UAE, disrupting services...
Chinese AI Turns Commercial Satellite Data Into Intelligence
Chinese AI Startup Watching US Military Assets in Middle East From Space So are a lot of actors... “The value of MizarVision’s output is not the raw satellite image. Any government can purchase commercial satellite passes. The value is the AI...
New Chip Lets Robots See in 4D by Tracking Distance and Speed Simultaneously
Researchers have created a silicon chip that integrates a 61,952‑pixel focal‑plane array capable of both emitting and receiving FMCW LiDAR signals, delivering 4D imaging that captures distance and speed simultaneously. The prototype generated detailed 3D point clouds at ranges from...
1M Context Is Now Generally Available for Opus 4.6 and Sonnet 4.6
Claude announced that Opus 4.6 and Sonnet 4.6 models now support a full 1 million‑token context window at regular pricing. The pricing remains $5/$25 per million tokens for Opus 4.6 and $3/$15 for Sonnet 4.6, eliminating any long‑context surcharge. Media limits have been expanded sixfold...
The DeepSeek Shock
After two years of a seemingly settled AI hierarchy dominated by OpenAI, Google, Anthropic, and Meta, Chinese lab DeepSeek abruptly released models that outperformed expectations. The models demonstrated performance comparable to the leading Western firms, and did so within a...
NYT: ‘Meta Delays Rollout of New AI Model After Performance Concerns’
Meta has postponed the launch of its next foundational AI model, code‑named Avocado, to at least May after internal benchmarks showed it lagging behind leading rivals such as Google Gemini 3.0, OpenAI and Anthropic. While Avocado outperforms Meta’s prior Llama 4 and...
Spotify Will Let You Edit Your Taste Profile to Control Your Recommendations
Spotify announced a beta feature that lets Premium users edit their Taste Profile, the algorithmic model behind Discover Weekly, Made For You playlists and Wrapped. The rollout begins in New Zealand, displaying listening data across music, podcasts and audiobooks in a...
The AI Race that Apple Is Winning
In this episode of Exponential View, host Azeem Azhar reexamines his view of Apple’s role in the AI race, highlighting a paradox: while Apple appears absent from AI model development and large‑scale data center investment, its hardware—especially high‑spec Mac Minis...
AI Needs Your Context, Not Just Faster Prompts
Most AI advice tells you to move faster, try more tools, learn more prompts I think the real problem is simpler than that: your AI doesn't know you It doesn't know your projects, your standards, or the decisions you've already made Better prompts...
OpenFold Consortium Announces Major OpenFold3 Update and Public Release of Training Data for Reproducible Biomolecular AI
The OpenFold Consortium unveiled OpenFold3’s major update, releasing the full training datasets, model weights, code, and evaluation scripts via AWS’s Registry of Open Data. The open‑source co‑folding system now includes a dedicated portal with onboarding documentation and a public support...
The Four Pillars of AI-Driven CX: Turning Technology Into Trust
Customer expectations are outpacing companies’ ability to adapt, with 70% of executives reporting a widening gap. AI alone isn’t enough; firms must embed it across four pillars—assisted agents, intelligent self‑service, operational automation, and adaptive AI orchestration—to turn CX into a...
Big Scores Mislead: Growth and Margin Matter
$NBIS has an R40 score of 445. $CRWV has an R40 score of 228. Higher is better, right? Not always. $CRWV's 228 = 168% growth + 60% margin. Both legs working. $NBIS's 445 = 457% growth + negative margin. One leg. The other is dragging...
Circle Skills Empower AI Agents for Stablecoin Integration
Our massive suite of infrastructure is now fully available to AI Agents and AI Agent builders with Circle Skills. Cursor, Claude Code, Codex, OpenClaw, come build the new internet financial system.
Manitoba Launches AI Consultations as It Weighs Tougher Privacy and Youth Protections
Manitoba’s government announced public consultations to shape future AI regulations. The process will address modernizing data‑privacy statutes and establishing responsible‑AI rules, including possible age limits and opt‑in consent for personal data. Consultations target Manitobans, Indigenous governments, educators, researchers, municipalities and...
Enterprise AI Fails Due to Training, Not Technology
AI failure in the enterprise isn't a technical problem; it's a training problem. You can't apply AI "evenly" and expect consistent results. If it's not invisible, it must be taught. No middle ground. 🚫 Learn more: https://t.co/KP9ClquPul #CIO #AI #Leadership #Upskilling
Academics Must Redefine Integrity for AI Era
How might academics best respond to AI in terms of academic integrity? Here's our full, energetic, and thoughtful conversation with Dr. Tricia Gallant: https://t.co/cJ5i6deVmu https://t.co/BnGPCbUwQJ
Rethinking Engineering Team Size in the AI Era
What's the right size and shape of an engineering team when AI tools are part of the equation? Maybe an uncomfortable question, but we need to talk about it. @dblockdotorg has thoughts: https://t.co/4kw5EziS8r
CIOs Must Assign Clear Ownership for AI Roadmaps
Does your AI roadmap doesn't name owners for data quality, deployment automation, and incident response? In the gen AI era, CIOs should make accountability more explicit #DevOps #DataGovernance #CIO https://t.co/7dcoLIKa0K
Join My Free Live AI Session Today at Noon
Doing a free live session at noon pst today (in about an hour). Going to share how I’ve been working with AI. You can still register and join here: https://t.co/wkYajstYeO
AI's Future Depends on Management Consulting Expertise
#AI Needs Management Consultants After All by @AllisonPohle @wsj Learn more: https://t.co/22WVGLJGTi #ArtificialIntelligence #MachineLearning #ML #DL https://t.co/BNYeCN5xJc
First Cloud Deploys NVIDIA Vera Rubin NVL72 for AI Validation
We’re the first cloud to bring up an NVIDIA Vera Rubin NVL72 system for validation, another big step in building the next generation of AI infrastructure with NVIDIA. https://t.co/apPyKh0HRK
Get $5K Grants for NYC/BOS AI Event Hackathons
giving $5k grants for NYC/BOS folks hosting events on AI agents / hackathons / etc for Tech Week in May/June:
ElevenLabs' Flows Unifies All Creative Media Tools
Honest take: ElevenLabs just killed every creative tool you're using. It's called Flows and it chains image, video, voice, music, and SFX into one visual pipeline. No tool-switching. No re-exports. No rebuilding from scratch. Here's how with real examples: https://t.co/4VY9L95RSD
How to Critically Evaluate Bold AI Announcements
Wading Through AI episode 2 is now up! In it, @DemetriSpanos and I discuss the recent Anthropic announcement about building a C compiler with "parallel Claudes", and how to properly assess AI announcements in general. https://t.co/OkHBjvRaqR
AI Governance Buzz Grows, Actual Impact Shrinks
AI Governance mentions up 53% but real influence dropped 22%. More people talking about it. Fewer organizations actually building it. Conference topic, not boardroom priority. That divergence tells you everything.
LLMs Aid Top Mathematicians but Remain Unreliable
LLMs are amazing. In math, they're clearly becoming useful to top mathematicians. They also struggle with reliability and not knowing what they don't know.