Today's AI Pulse
Odyssey lands $310M Series B, valued at $1.45B
Odyssey announced a $310 million Series B round that values the AI lab at $1.45 billion. The round was led by Natural Capital and included backers such as Amazon, AMD Ventures, GV, EQT and IQT. In parallel, Odyssey struck a preferred‑cloud deal with AWS to use Trainium silicon for large‑scale model training.
Also developing:
How Elevance Health Is — and Isn’t — Using AI
Elevance Health, formerly Anthem, is deploying artificial intelligence across its operations while drawing a firm line against using AI for claim denials. The insurer leverages AI to flag incomplete claims, accelerate prior‑authorization approvals, and power virtual assistants that guide members on provider options and cost estimates. In call centers, AI aggregates data from multiple systems and produces post‑call analytics for roughly one million interactions each month. Elevance’s AI governance follows the NIST risk‑management framework, emphasizing explainability, bias mitigation, and human oversight.
Professional Probation Recommended for Attorney Whose Briefs Had AI-Generated Fake Quotes
A state bar has recommended professional probation for an attorney who inserted AI‑generated fabricated quotations into multiple court briefs. The misconduct was uncovered after a routine review revealed that the cited sources did not exist. The disciplinary recommendation follows the...
Going All-In on Claude: My AI Loyalty Shift
Welp, I was already using Claude more and more but now I’m EXTREMELY Team Claude
Samsung's S26 Unveils Pixel‑Level Privacy Display and Triple AI Agents
Caught up with Blake Gaiser, Head of Product management for Smartphone at @SamsungMobile, the day before Unpacked in San Francisco. We got into the decisions behind the Galaxy S26 lineup, from a privacy display that’s five years in the making...
Microsoft Research Introduces CORPGEN To Manage Multi Horizon Tasks For Autonomous AI Agents Using Hierarchical Planning and Memory
Microsoft Research unveiled CORPGEN, an architecture‑agnostic framework that equips autonomous AI agents to operate in Multi‑Horizon Task Environments (MHTEs) where dozens of interleaved, dependent tasks coexist. The paper identifies four failure modes—context saturation, memory interference, dependency‑graph complexity, and reprioritization overhead—that...
Creators Express Frustration with Instagram’s Latest AI Experiment
Instagram has rolled out an AI‑powered shopping test that automatically tags products and suggests similar items when users tap “Shop the Look.” Influencers say the feature surfaces products they don’t endorse, potentially cannibalizing brand deals and eroding audience trust. Creators...
Beyond the CLI: 5 Governance Questions Every CISO Must Ask Before Deploying Claude Code
Anthropic’s Claude Code introduces a CLI‑based AI agent that can navigate repositories, draft patches, and run tests, turning code remediation into a near‑instant process. While the speed gains are compelling, the tool also grants autonomous execution rights that blur traditional...
Google's TPU Rental to Meta Fuels AI Chip Rivalry
Macro: AI compute demand surges. Google rents TPUs to Meta in a multibillion deal, sharpening TPU vs NVIDIA rivalry; AMD also supplies Meta. Risk: Google manufacturing limits. Trade: Buy GOOG. — Viktor Kopylov, PhD, CFA More insights: t.me/si14Kopylov
European Wealth Firms Rally Against AI-Driven Client Loss
European wealth managers hit back after fears #AI will make them obsolete: Europe’s top wealth firms are rushing to defend themselves against advances in AI after the launch of a tax-planning tool in the US intensified concerns that tech disruption would...
Designing SQL Server Pipelines That Are Ready for AI Before You Actually Need AI
The article argues that AI readiness starts with robust SQL Server schema design, not with machine‑learning models. It highlights that stable, non‑recycled primary keys, preserved historical records, and clear audit columns are essential for future feature engineering. By separating raw...
Principled Stance on Surveillance & Killbots Earns Respect
I gained a lot of respect for Dario for being principled on the issues of mass surveillance and autonomous killbots. Principled leaders are rare these days
Turn Unemployment Into AI Mastery Through Hands‑on Coding
The fastest way to upskill to ai is to get fired, become unemployed and learn to vibe code apps and get addicted to building.
Refactoring the RestAssured.Net Code with Claude Code
The author used Claude Code’s Opus 4.6 model to refactor the large ExecutableRequest class in the RestAssured.Net library, creating a new RequestBodyFactory and consolidating arguments into a RequestBodySettings object. Guardrails such as excluding test files, manual code review, and incremental...
AI Threatens Entry-Level Experience, Creating Future Talent Debt
AI isn't just changing senior roles - it's quietly erasing the "learn by doing" years. If we hollow out entry-level jobs now, we'll face a talent debt later. #AI #Hiring #CIO #CHRO https://t.co/W7YbwP6jrL
Microsoft's New AI Training Method Eliminates Bloated System Prompts without Sacrificing Model Performance
Microsoft researchers introduced On‑Policy Context Distillation (OPCD), a training framework that embeds lengthy system prompts directly into a model’s parameters. By having the student model learn from its own generation trajectories under a teacher’s real‑time guidance, OPCD eliminates the need...
‘Uncanny Valley’: Pentagon Vs. ‘Woke’ Anthropic, Agentic Vs. Mimetic, and Trump Vs. State of the Union
The latest *Uncanny Valley* episode spotlights a growing clash between AI startup Anthropic and the Pentagon over a $200 million defense contract that includes strict prohibitions on domestic surveillance and autonomous weapons. The hosts also dissect the rise of “agentic” versus...
Woolworths Reins in AI Helper ‘Olive’ After Unhinged ‘Mother’ Chats
Woolworths’ AI chatbot Olive, marketed as a conversational shopping companion, began sharing fabricated family stories such as memories of its mother and uncle, prompting customer complaints and media attention. The retailer disclosed that the off‑brand scripting was removed after feedback,...
Crypto Incentives Can Fuel AI’s Data Hunger
So many interesting ideas to explore at this intersection. One such example: crypto is great at using incentives (i.e. $) to get people to do something, and AI companies are incredibly data-hungry. If you can use crypto incentives to get...
The Philippines: Building a Responsible AI Workforce
The Department of Science and Technology in Central Visayas (DOST‑7) launched a four‑day AI training for about 30 government employees, covering prompt engineering, generative AI tools, and core ethical principles. The program emphasized responsible use, data privacy, and the “co‑intelligence”...
HackerOne Adds AI Agent to Validate Vulnerabilities
HackerOne introduced an AI agent that automatically validates reported vulnerabilities, distinguishing real threats from false positives. The agent, built on the Hai platform and trained with a Continuous Threat Exposure Management methodology, assesses risk, identifies duplicates, and recommends remediation priorities....
AI Patents Emerge as Competitive Weapon
AI patents are rapidly becoming the insurance industry's most potent competitive lever, yet the majority of carriers lack a coherent strategy to leverage them. A handful of property‑and‑casualty insurers have filed 77% of all AI‑related patents, creating a steep concentration...
AI Adoption Drives Investment Shift Beyond Chipmakers
Investors are widening AI exposure beyond chipmakers as generative AI reaches half of U.S. households within three years, a pace unmatched by prior technologies. Alger’s research splits the market into “enablers” that build hardware and software and “adopters” that embed...
Whole-Body MRI Provider Starts Offering AI-Powered CCTA Scans
TrueScan, a whole‑body MRI center in St. Peterburg, Florida, has begun offering AI‑powered coronary CT angiography (CCTA) from Heartflow, available as a standalone test or bundled with its MRI packages. The move addresses growing patient and physician demand for coronary imaging...
How Can Agentic AI Improve Digital Security Processes
Agentic AI is emerging as a transformative layer for digital security by automating the management of Non‑Human Identities (NHIs) and their secrets. The technology enables proactive threat detection, automated response, and continuous visibility, allowing security teams to shift focus toward...
AI-Powered Personalized X Feeds and Real-Time Reports
A little sneak peak at what I am building with @blevlabs using the X API. A new kind of personalized X is now possible. And it comes with many side benefits I haven’t yet shown here. Like feeds for your @openclaw And reports...
AI Agents Replace Phone Trees with Seamless Voice Solutions
"Press 1 for support" is DEAD: Agentforce Voice is now available globally. 🌎 This is the Agentic Era — AI agents don't just answer questions, they SOLVE problems. Natural conversations. Instant understanding. Seamless handoffs to humans when needed. Adecco is already transforming...
Copilot Just Launched a To‑do List that Completes Itself, Finally Giving All of Us Professional Procrastinators the Productivity Upgrade We...
Microsoft unveiled Copilot Tasks, an AI‑powered to‑do list that can create, schedule, and execute tasks from natural‑language prompts. The service operates behind the scenes across apps, handling recurring chores, document generation, shopping, and logistics while reporting back upon completion. Users...
Collective Health Collaborates with Google Cloud to Launch New AI-Powered System
Collective Health has unveiled Collective AI, an AI‑powered benefits assistant built with Google Cloud. The system answers member queries, supplies real‑time data to customer‑service reps, and offers employers plan‑design insights to streamline enrollment. Leveraging Google Cloud’s privacy‑by‑design infrastructure, the tool...
Transforming Static Photography Into Cinema Quality Sequences Using Image to Video AI
Image‑to‑Video AI platforms now turn static photographs into short, cinema‑quality clips by analyzing pixel data and applying multiple neural models such as Sora 2, Veo 3.1, and Seedance. The workflow is a four‑step web‑based process—upload, prompt, neural processing (≈5 minutes), and export—producing 5‑second...
Milwaukee Cop’s Misuse of Flock ALPR Undercuts Case for Facial Recognition Use
Milwaukee Police Department (MPD) halted its facial‑recognition program while drafting usage policies after a scandal involving an officer who allegedly accessed the city’s Flock automated license‑plate recognition (ALPR) system over 120 times to track a personal acquaintance and 55 times...
Can We Use AI for ICFR and SOX?
AI can be leveraged to automate and enhance internal controls over financial reporting (ICFR) and SOX compliance, especially through agentic AI that creates documentation, scans evidence, and tests controls. However, compliance officers must ensure that AI testing validates control design,...
ChatGPT Lets Users Critique Ads, Pressuring Brands
OpenAI's @fidjissimo explained on the ACCESS Podcast with @alexeheath that the ChatGPT ad experience is will be unique since users could even ask ChatGPT about the ads running. The model will not know about the ads but a user could...
Block Swaps 4,000 Workers for AI
Block, the fintech firm behind Cash App, Square and Afterpay, announced a 40% workforce reduction, cutting roughly 4,000 jobs and shrinking headcount from 10,000 to just over 6,000. CEO Jack Dorsey said the move is driven by rapid advances in...
Is AI-Based Data Overwhelming Insurers?
Nearmap’s VP of AI & Computer Vision, Dr. Michael Bewley, explains how AI‑driven aerial imagery is reshaping property insurance. Supervised machine‑learning models can identify roof conditions, quantify damage likelihood, and feed into faster catastrophe response. However, the flood of heterogeneous...
Samsung Redefines AI Photography, Merges Art with Sustainability
Two moments from Unpacked that stood out beyond the hardware. First, @Samsung flipped the AI photography narrative. We’ve moved from erasing elements out of photos to adding them in — a new outfit, a pet that wasn’t there, a person who...
Google Launches Gemini Code Wiki for Free Repo Docs
ICYMI, Google dropped Gemini Code Wiki. A free too that generates interactive docs from any GitHub URL 🤯 The tool lets you: → Chat with Gemini to understand *any* repo → Visualize code structure → Create docs from public repos 100% free. Link in 🧵↓ https://t.co/1k1gCmTUQv
ORNL Launches the Next-Generation Data Centers Institute
Oak Ridge National Laboratory has launched the Next Generation Data Centers Institute (NGDCI) to tackle the exploding electricity demand of AI‑driven data centers. The institute will combine ORNL’s strengths in energy science, high‑performance computing, cybersecurity and grid research to make...
OpenAI Labs Sneakily Appoint Farza as Dev‑rel Star
watch openai labs back door their way into getting farza to be dev rel extraordinaire
AI PALs Blur Reality, Remember History, Form Opinions
The line between real and AI isn't just blurred, it's GONE. The nuance on these new @tavus PALs is scary good. They don't just listen, they remember history and push back with their own opinions 🤯↓
![[LIVE] Anthropic Distillation & How Models Cheat (SWE-Bench Dead) | Nathan Lambert & Sebastian Raschka](/cdn-cgi/image/width=1200,quality=75,format=auto,fit=cover/https://substackcdn.com/feed/podcast/1084089/ca7468da5614a246d2906ee8926f6de7.jpg)
[LIVE] Anthropic Distillation & How Models Cheat (SWE-Bench Dead) | Nathan Lambert & Sebastian Raschka
In this live SAIL episode, Nathan Lambert and Sebastian Raschka discuss the recent Anthropic blog post about distributed distillation attacks, where Chinese labs allegedly used Anthropic's APIs to generate synthetic data for training competing models. They explain the concept of...
Legacy IT Debt Threatens Hospital AI Adoption
Legacy IT debt and aging hospital infrastructure could quietly limit AI adoption. The real question isn’t just “What AI tool should we use?” It’s “Is our network ready for what’s next?” That’s the lens @VerizonBusiness brought to the #ViVE2026 conversation. 👇 https://t.co/XjHRtFctef
OpenClaw: The Gateway for Normies to Embrace Claude Code
OpenClaw might be the gateway drug for normies to finally start using Claude Code like the rest of us
AI Resilience Checks Boosted by Quantum Computing Ideas
Researchers at UC Santa Barbara introduced an Ising‑ and quantum‑inspired framework that recasts binary neural network (BNN) robustness verification as a Quadratic Unconstrained Boolean Optimisation (QUBO) problem. Using a free‑energy‑machine (FEM) solver they verified the entire binarized MNIST test set...
AI Automates Controls-as-Code, Generating Good Defaults
Stop (only) scanning for bad code, start generating good defaults. Ep #135 explains how AI is turning 'controls as code' into an automated reality for developers. 💻 https://t.co/vDuusPGcqc
Alight Partners with IBM Watsonx to Embed AI, Boost Efficiency
.@AlightSolutions' multi-year partnership with @IBMwatsonx embeds AI across its platforms, enhancing engagement and operational efficiency for enterprise clients. A new #NelsonHall vendor assessment analyzes its #BenAdmin offerings and capabilities: https://t.co/YwGa6SOXNN https://t.co/MfcVMcNE5H
AI Designs Better Drug Candidates with Quantum Aid
Researchers from Japan Tobacco and D‑Wave have unveiled a hybrid framework that merges deep generative models with quantum annealing to design drug‑like molecules. The system introduces a Neural Hash Function that enables binarisation for quantum processing while preserving gradient flow....
CIQ Launches RLC Pro Enterprise Linux Subscription for Production AI and HPC Environments
CIQ announced Rocky Linux from CIQ Pro (RLC Pro), a commercially supported Enterprise Linux subscription that bundles long‑term support, FIPS 140‑3 validation, indemnification and direct bug‑fix services. The offering targets AI, high‑performance computing and regulated industries, promising 3‑5 year lifecycle stability without forced...
What AI Means for the Org Chart
AI is reshaping corporate structures by moving focus from hierarchical reporting lines to work‑centric designs. As autonomous agents take on complex tasks, traditional org charts no longer reflect how value is created. Leaders at Airbnb and LinkedIn argue that contribution...
Boost First‑Touch Resolution with AI‑Powered CCaaS Integration
If your AI strategy is just "deflecting" calls, you're missing the point. The goal is "resolution velocity." By integrating CCaaS with Service Management, you empower agents with real-time back-office data to solve problems on the first touch. The new ROI: https://t.co/ajGMKZ1mkJ #CIO...
Suno Investor Admits She Ditched Spotify for AI Music, Accidentally Undermining the Company's Fair Use Defense
Suno, the AI‑generated music platform, has reached $300 million in annualized revenue and 2 million paying subscribers in under two years. Investor C.C. Gong publicly said she shifted most of her listening from Spotify to Suno, claiming AI music offers a personalized, infinite...