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
Creating Time for AI Experimentation Key to ROI
Atlassian’s recent internal research shows only 24% of AI projects target team outcomes, even though knowledge workers spend about 80% of their time collaborating. The company’s HR vice‑president Alicia Lenart argues that a culture of experimentation is essential to capture AI’s hidden ROI potential. By embedding AI into workflows and focusing on collective productivity, firms can move beyond individual‑centric metrics. This shift promises larger, more sustainable returns on AI investments across the enterprise.
Anthropic Secures $65B Series H, Valuation Hits $965B, Overtaking OpenAI
Anthropic closed a $65 billion Series H round, lifting its post‑money valuation to $965 billion and briefly overtaking OpenAI as the most valuable AI startup. The capital, led by Altimeter Capital, Dragoneer, Greenoaks and Sequoia, will fund safety research, compute expansion and enterprise...
Company Directors Face Rising Legal Risk and Responsibility over AI
Chief Justice Andrew Bell warned that artificial intelligence is now a core component of corporate governance in Australia, and directors who fail to understand or oversee it may breach existing fiduciary duties. While the Corporations Act remains technology‑neutral, the "black‑box"...
How Do You Evaluate An AI Agent? (The Agents Season, Episode 7)
The latest episode of The Agents podcast explores the thorny problem of evaluating AI agents, highlighting how agents can appear successful while silently failing or entering endless loops. It examines why traditional performance metrics often miss subtle errors and proposes...
Apple's AI Push Could Bypass Siri, Redefining Marketing on iPhone
Bank of America analysts say Apple’s next AI showcase will likely treat Siri as an execution layer rather than a standalone chatbot, positioning the iPhone as the primary endpoint for agentic AI. The move could unlock $15‑$30 billion in incremental revenue...
Kirkland & Ellis Allocates $500 Million to Build Proprietary AI Platform
Kirkland & Ellis announced a $500 million, multi‑year commitment to create a proprietary artificial‑intelligence platform for its global practice, beginning with a $100 million infusion in 2026. The move signals a shift from third‑party tools to in‑house solutions aimed at enhancing contract...
AI Enables Parkinson’s Musician to Finish Album Amid Industry Lawsuits
Samuel Smith, a London‑based Americana singer, turned to AI music generators Suno and Udio to finish his second album “The Art of Letting Go” after Parkinson’s disease robbed him of guitar dexterity. The move underscores AI’s therapeutic potential while the...
Anthropic's Opus 4.8 Upgrade Boosts Claude Efficiency but Raises Cost Risks
Anthropic released Opus 4.8 on Thursday, introducing dynamic workflows and an effort‑dial that promise smarter, faster Claude responses for business‑critical workloads. The upgrade keeps the headline price unchanged but triples the speed of fast mode, while also exposing enterprises to...
MeMo Framework Lifts LLM Efficiency 26% without Retraining, Easing AI‑driven DevOps
A team of university researchers introduced MeMo, a modular framework that boosts large language model (LLM) efficiency by 26% without any model retraining. By offloading new information to a dedicated “memory” model, MeMo sidesteps costly fine‑tuning and the fragility of...
Dell's Record Q1 Boosts AI Server Revenue, Sends Palantir Up 10%
Dell Technologies reported fiscal Q1 revenue of $43.84 billion, an 88% year‑over‑year jump, and AI‑optimized server sales that surged 757% to $16.1 billion. The results lifted Dell’s share price 32.8% and triggered a roughly 10% rally in Palantir stock, underscoring the market...
Clouted Raises $7M Seed to Power AI‑Driven Viral Reels for Creators
Clouted, an LA‑based AI startup, announced a $7 million seed round led by Slow Ventures to expand its “Distribution Intelligence” platform that automates short‑form video testing and publishing. The funding backs a product that claims to turn every campaign into a...
Robinhood Launches AI Agents for Automated Trading and Credit Card Purchases
Robinhood unveiled Agentic Trading and an Agentic Credit Card on May 27, 2026, giving its 27 million customers AI‑driven bots that can execute stock trades and shop with a virtual credit card. The beta, limited to equities, marks the first mainstream retail brokerage...
AI Broke Your Experimentation Program. Here’s How to Fix It.
AI has turned experiment ideation into a near‑instant process, making raw test velocity a poor health metric. Teams now need disciplined idea filtering, trustworthy data analysis, and the ability to experiment beyond traditional UI elements. Modern, warehouse‑native platforms unify experiment...
Bookmark This: The Free AI Learning Path for 2026
The post presents a free, structured AI learning path for 2026 that organizes content by skill layers rather than by provider. It starts with AI fundamentals, then progresses through large language models, prompting, agents, automation, Claude, Model Context Protocol, and...
CEO Interview with Vivek Vishwakarma of ThirdAI Automation
ThirdAI Automation, led by CEO Vivek Vishwakarma, builds causal AI for semiconductor equipment and secured a $3 million seed round. Its platform shrinks diagnostic cycles from eight hours to roughly twelve minutes by running on standard CPUs inside air‑gapped fab environments....
Clarivate Launches IPOne AI Platform to Transform IP Research and Workflow
Clarivate announced IPOne, an AI‑powered platform that unifies its Derwent patents, Darts litigation records and CompuMark trademark data into generative‑AI agents. The launch aims to shift the company from a reference‑data provider to a daily decision‑support tool for corporate IP...
CIQ Extends Fuzzball Orchestration to Five Clouds and On‑Premises
CIQ announced that its Fuzzball orchestration platform now runs across five public clouds—CoreWeave, AWS, Google Cloud, Oracle Cloud and Microsoft Azure—and on‑premises clusters. The expansion lets AI and high‑performance computing teams define a workflow once and execute it anywhere, cutting...
Insurance Analytics Market Projected to Hit $72.39 B by 2035, Driven by AI and Cloud Adoption
Research firm SNS Insider forecasts the global insurance analytics market will grow to $72.39 billion by 2035, up from $9.20 billion in 2025. The projection rests on rapid adoption of cloud computing, big data and artificial intelligence across insurers and insurtech firms,...
How Does Open Source AI Software Compare With Leading Commercial AI Software?
The Open Source Initiative’s 2024 definition now clarifies what qualifies as open‑source AI, covering frameworks, model libraries, inference engines, agent tools, and local runners. Commercial AI suites—such as OpenAI’s GPT‑5.5, Anthropic’s Claude Opus 4.8, and Google’s Gemini—offer managed services, compliance tooling,...
Shopify Unveils Storefront AI, a Native Personalization Engine for Plus Merchants
At Unite 2026 in Toronto, Shopify introduced Storefront AI, an infrastructure‑level personalization engine baked into its Storefront API and already in beta for 300 Plus merchants. The feature promises real‑time product reordering and copy rewriting without third‑party apps, challenging a...
Map Shows 4,300+ AI Data Centers Near Lake Mead as Reservoir Hits Historic Low
A new map reveals more than 4,300 AI‑focused data centers within the Lake Mead watershed, including 70 in Nevada. With the reservoir at just 30% full and combined water use of nearby facilities exceeding 716 million gallons a year, residents and...
Photon-Driven Synapse Advances Low-Power Neuromorphic Systems
Researchers have created a fully optical artificial synapse using a rare‑earth‑doped crystal that stores and processes information with light alone, eliminating electrical conversion steps. The device demonstrates both paired‑pulse facilitation under UV light and paired‑pulse depression under near‑infrared light, mimicking...
AI Darwinism vs Leadership: Who Will Survive?
The race between AI Darwinism and leadership decides who evolves and who doesn't... Please read and subscribe 🙏 https://briansolis.substack.com/p/the-new-ai-divide-companies-that
AI-Generated Visual Novel Game Is On Steam For $100
An indie visual novel called Blood in the Ice is listed on Steam for $100, dropping to $59.99 during sales. The game’s assets, including all images and store screenshots, are almost entirely AI‑generated, a fact disclosed in its description. Developer...
Claude Mythos Exposed a Hard Truth: Your Enterprise Patching Process Is Way Too Slow
Anthropic’s Claude Mythos preview proved AI can autonomously discover thousands of zero‑day vulnerabilities, collapsing exploitation timelines to hours. Recent CVEs such as Langflow and Marimo were weaponized within 20 hours and under 10 hours of disclosure, far faster than the...
Google's AI-Focused Search Drives 3x Traffic Spike for DuckDuckGo's 'No AI' Feature
DuckDuckGo reported that traffic to its “No AI” search page has tripled since Google unveiled its AI‑driven Gemini 3.5 Flash overhaul at I/O. The surge began around May 19 and has held steady, with daily visits about 84 % above the prior average. In the...
NASA Unveils Hand‑Sized Processor That Boosts Spacecraft Computing 100‑Fold
NASA announced a hand‑sized, radiation‑hard system‑on‑a‑chip that offers 100 times the computing capacity of existing spacecraft computers. Developed with Microchip Technology, the processor aims to enable AI‑driven autonomy on future lunar, Martian and deep‑space probes.
MISUMI Group Deploys $1 B to Launch MISUMI Americas, Boost AI‑Driven Manufacturing
MISUMI Group announced a $1 billion investment to create MISUMI Americas, integrating Fictiv’s AI‑powered digital manufacturing platform with its 60‑year legacy of precision components. The move, led by newly appointed U.S. CEO Dave Evans, aims to turn static supply chains into...
Mumbai Deploys AI Platform CivitTwin to Speed Building Approvals
Mumbai’s municipal corporation has launched CivitTwin, an artificial‑intelligence platform that automates early‑stage scrutiny of construction applications. The system is designed to identify missing approvals and regulatory gaps, promising faster, more transparent building clearances in the city’s dense real‑estate market.
Delaware Chancery Court Orders AI Chat Logs Preserved in Litigation Holds
The Delaware Court of Chancery ruled in March 2026 that AI chat logs are discoverable evidence, requiring them to be included in litigation holds. The decision in Fortis Advisors LLC v. Krafton Inc. compels law firms and corporate legal departments...
SoftBank Pledges €75 Bn ($87 Bn) for 5 GW AI Data Center Build‑out in France
SoftBank Group announced a €75 bn ($87 bn) investment to develop 5 GW of AI data center capacity in France, with an initial 3.1 GW rollout in Hauts‑de‑France by 2031. The plan, backed by French officials and partners like Schneider Electric and EDF, aims...
Guidewire CEO Says Modern Core Platforms Are Crucial for Agentic AI in Insurance
Guidewire President John Mullen told industry leaders that insurers must upgrade to modern core policy, claims and billing systems before deploying agentic AI that can act autonomously. He warned that without a reliable foundation, AI‑driven decisions could jeopardize auditability and...
Terzo AI Founder Turns Contract‑Copy‑Paste Routine Into $100 M Startup
Brandon Card, a former executive at Microsoft, Oracle and IBM, founded Terzo AI after a frustrating contract‑copy‑paste exercise. The company has raised more than $40 million and now serves the world’s largest enterprises, turning a $100 million deal pain point into a...
OpenAI Robotics Hiring Engineers to Build Future Robots
OpenAI Robotics is hiring, looking for exceptional full-stack hardware, ops, systems, and ML engineers to help us program and manufacture robots that are useful for society. AI should be able to help people in the physical world. In the short term,...
From Reactive Operations to Autonomous Infrastructure: What IT Leaders Must Do Next
IT leaders are transitioning from reactive monitoring to autonomous infrastructure, where AI agents not only alert but also remediate issues. While enthusiasm is high, only 5% of surveyed professionals say AI is core to their operations, highlighting a gap between...
Tempus Unveils the Next-Generation of Lens, Expanding Its Agentic AI Platform for Oncology Drug Development
Tempus AI launched the next‑generation Lens, an agentic AI platform that unifies its multimodal real‑world oncology data, high‑performance computing and validated AI agents. The cloud‑based system lets drug developers create research plans in plain language, execute analyses on more than...
Claude's Honest Haul, Figure Leaves the Line, and SpaceX's Big Bang
Anthropic unveiled Opus 4.8, a sharper LLM that cuts code‑error slips by four times while keeping the same price as 4.7, and simultaneously closed a $65 billion Series H round that values the company at $965 billion. Robotics firm Figure signed a commercial deal...
1-Bit Bonsai Image 4B Image Generation for Local Devices
PrismML unveiled Bonsai Image 4B, a pair of ultra‑compact diffusion models—1‑bit and ternary—that compress transformer weights to under 1 GB. The binary variant achieves an 8.3× reduction, while the ternary version retains 95 % of FLUX.2 Klein 4 B quality with a 6.4× size cut....
I Was Wasting 31% of My Context Window Before I Fixed My MCP Setup
The author discovered that loading every MCP tool definition ate roughly 31% of Claude’s usable context window, a problem mirrored in GPT‑4o where tool schemas consume a similar share. This hidden token tax slows agents, increases wrong‑tool calls, and inflates...
ByteDance’s TokenMixer-Large: Scaling Ranking Models
ByteDance unveiled TokenMixer-Large, a deep ranking model that overcomes the gradient‑vanishing problem of its predecessor RankMixer. The architecture introduces a symmetric Mixing‑Reverting block that keeps token dimensions aligned, enabling very deep networks. By stripping away memory‑bound operators and relying almost...
AI‑Native Banking OS Shifts Focus From Accounts to Customers
The End of Account-Centric Banking: Jouk Pleiter on Backbase's AI-Native Banking OS See the full episode here: 👉 https://t.co/cYBd83keTh What does an AI-native bank actually look like? Jouk Pleiter, founder and CEO of Backbase, argues that adding AI to a fragmented banking...
Joyson Unveils Human‑Like AI Robot Head Module
Joyson’s New #AI #Robot Head Brings Human-Like Motion, Expression, and Interaction Into One Smart Module by @XRoboHub #Robotics #Engineering #ArtificialIntelligence #Innovation #Technology https://t.co/UCrl2TYNEA
Apple Seeks to Disrupt the Glasses Market the Way It Did With Watches
Apple is preparing a new line of iPhone‑connected smart glasses to challenge Meta’s Ray‑Ban Gen 2 AI eyewear. The upcoming iOS 27 Siri app will enable seamless data sync across devices via iCloud, reinforcing the glasses’ integration with Apple’s ecosystem. Simultaneously, Apple...
Mass Effect and Metal Gear Actor Jennifer Hale Calls on Studios to Take Responsibility for Their AI Use: 'Ain't Nobody...
Voice actress Jennifer Hale called on game studios to take responsibility for their use of generative AI, demanding consent, control, and compensation for actors whose performances are used as training data. She emphasized that AI is a tool and that...
Netflix Acquires AI Startup and Expands Wonka Partnerships to Trim Costs and Drive Growth
Netflix has purchased Ben Affleck‑backed AI startup InterPositive and sealed new licensing agreements with Ferrero and Moose Toys for the Wonka universe. The moves aim to lower production expenses through automation while creating fresh merchandising revenue streams.
Relativity and Salesforce Deploy Anthropic’s Claude API, Boosting E‑Discovery and Code Automation
Relativity integrated Anthropic’s Claude Compliance API into RelativityOne, and Salesforce announced an 18‑times faster API migration using Claude Code agents. Both moves underscore a surge in large‑language‑model adoption across core legal‑tech workflows.
Features of SAEs Are Universal - but only up to an Unknown Random Rotation
Researchers examined Sparse Autoencoders (SAEs) trained on identical transformer architectures but different random seeds. Although decoder‑column cosine similarities exceed 0.9, applying one model's SAE to another’s activations catastrophically fails, producing negative explained variance. By fitting a single orthogonal Procrustes rotation,...
Anthropic Adds Parallel Sub‑agents to Claude Code, Speeding DevOps Pipelines
Anthropic released a dynamic‑workflow mode for its Claude Code AI, allowing large coding jobs to be split into parallel sub‑agents that generate and validate code in a generator‑validator loop. The feature runs unattended for 30–50 minutes, offering a new tool...
AI Data Center Trial Cuts UK Grid Power Use 35% in Real Time
Emerald AI, National Grid, Nvidia and Nebius proved that an AI data centre in London can reduce its electricity draw by about 35% in 30 seconds and hold the reduction for up to 10 hours, all while keeping AI workloads...
Meta to Test AI‑Powered Pendant and Roll Out Four Smart‑Glasses Models by 2026
Meta disclosed an internal memo outlining a test of an AI‑enabled pendant that records speech for on‑the‑fly summaries, and a roadmap to release four smart‑glasses models by the end of 2026. The moves accompany a $3.99 consumer subscription and a...