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
Boosting Crop Yield Accuracy with MHCNN-LSTM-MHA Model
Researchers published a new interpretable deep‑learning model, MHCNN‑LSTM‑MHA, that fuses multi‑head convolutional networks, long short‑term memory units, and multi‑head attention to predict crop yields. The hybrid architecture extracts spatial features from satellite and soil data, captures temporal weather patterns, and adaptively weights influential variables. Benchmarked across diverse crops and climates, it consistently outperforms existing deep‑learning and statistical models while tolerating noisy, incomplete inputs. Its scalability makes it ready for real‑time precision‑agriculture platforms.
Deep Dive Reveals Real Barriers to Legal AI Adoption
Extremely thorough and clear piece about AI adoption in the legal industry, and what the actual barriers and challenges are. Way more writing about AI should be exactly like this.
Can Large Language Models Identify Novel Threats? Part 1: Mirror Life and the Classification Gap
The article examines whether large language models (LLMs) can refuse harmful uplift requests about emerging threats that lack formal classification, using mirror life—a synthetic, chirally inverted organism created in 2022—as a case study. It highlights a gap between rapid scientific...
How Should We Update on AI-Enabled Coups Post-Mythos?
Anthropic’s Claude Mythos, deemed too dangerous for public release, can autonomously discover and exploit thousands of software vulnerabilities, turning zero‑day attacks into an industrial‑scale process. The model’s ability to out‑code most humans and expose a decades‑old flaw in a leading...
Out-of-Context Reasoning (OOCR) in LLMs: A Short Primer and Reading List
Out‑of‑Context Reasoning (OOCR) refers to LLMs reaching conclusions that require multi‑step logic without any intermediate steps appearing in the prompt. The primer defines OOCR, contrasts it with in‑context (CoT) reasoning, and showcases examples such as 2‑hop factual queries and inductive...
Accenture Backs Aera to Accelerate AI-Driven Supply‑chain Decisions
Accenture has invested in Aera Technology through its Accenture Ventures arm, pairing Aera's agentic decision intelligence with Accenture's AI‑enabled supply‑chain services. The partnership targets complex, global supply chains across consumer goods, high‑tech, life sciences, mining and oil & gas, where...
Abridge Hires Former Slack and Notion Exec San Oo as CTO to Scale AI‑driven Clinical Platform
Abridge announced San Oo, a former senior engineering leader at Slack and Notion, as its new chief technology officer. The hire is intended to accelerate the company’s AI platform, which already serves more than 270 major U.S. health systems and...
BioVigil Names Brad Ryba CTO to Boost AI‑enabled Hand‑hygiene Platform
BioVigil announced the appointment of Brad Ryba as chief technology officer, tasking him with scaling the company’s AI‑powered hand‑hygiene platform. Ryba brings more than 25 years of technology leadership across SaaS, healthcare, and telecom, positioning BioVigil to deepen data integration...
Siemens Launches AI‑Enabled Managed Building Service for Real‑Estate Operators
Siemens Smart Infrastructure introduced Asset Performance Advanced, an AI‑enabled managed building service that turns sensor data into automated maintenance actions. The solution combines predictive analytics, prescriptive AI and workflow automation to cut costs and improve occupant comfort for commercial real‑estate...
Harvey Teams with DeepJudge to Fuse AI Reasoning with Firm Knowledge
Harvey has joined forces with DeepJudge to integrate its legal AI platform with DeepJudge's institutional knowledge system, creating a unified workflow that surfaces a firm’s own historical documents during AI‑assisted research, drafting and decision‑making. The partnership targets the long‑standing problem...
Alibaba Launches Zhenwu M890 AI Accelerator and Panjiu AL128 Supernode Server
Alibaba's semiconductor arm T-Head announced the Zhenwu M890 AI accelerator and the Panjiu AL128 rack‑scale server, touting three‑fold performance gains over its predecessor. The company disclosed it has fabricated only 560,000 chips to date, highlighting capacity constraints as it battles...
Corsair Unveils Corsair Pro AI Workstations and Servers for Enterprise Data Centers
Corsair announced the Corsair Pro lineup of AI‑focused workstations and servers, aiming to serve data‑center and enterprise AI developers. The products expand Corsair's hardware portfolio beyond consumer PCs, signaling a strategic push into the fast‑growing AI infrastructure market.
Louisiana Deploys AI‑Powered Coding Platform in 10 Parishes' High Schools
Bossier Parish Community College and the Louisiana Department of Education launched a pilot that trains 21 high‑school teachers across ten parishes on the AI‑driven Kira Learning platform. The program aims to scale computer‑programming instruction statewide, with the first student classes...
GitLab 19.0 Launches AI‑Driven Workflows, Secrets Manager Beta and Self‑Hosted Model Support
GitLab released version 19.0, embedding AI‑driven merge request workflows, a public‑beta Secrets Manager and support for self‑hosted open‑source AI models. The upgrade aims to reduce handoffs, improve pipeline visibility and secure software supply chains for DevOps teams.
Dell Shares Jump 14.7% on AI Factory Surge and Analyst Target Raises
Dell Technologies' shares surged 14.7% to about $289 after strong AI‑focused demand and a wave of bullish analyst price targets, including Mizuho’s $300 forecast. The rally underscores the market’s confidence in Dell’s AI Factory strategy and its expanding enterprise infrastructure...
CMU and Cleveland Clinic Launch CMR-CLIP AI, Boosting Cardiac MRI Accuracy to 99%
Carnegie Mellon University and Cleveland Clinic introduced CMR-CLIP, an AI system that outperforms generic models by more than 35% and achieves up to 99% accuracy in cardiac MRI interpretation. Trained on over 13,000 de‑identified scans, the tool aims to cut...
National Bank of Canada Deploys Sardine AI Platform, Leads $25 Million Funding Round
National Bank of Canada signed a multi‑year contract to embed Sardine’s agentic AI risk platform across its retail, commercial and wealth divisions, while also leading a $25 million Series C extension that lifts the startup’s total capital to $170 million. The deployment aims...
OneStream Unveils Forward Finance Blueprint to Accelerate AI‑Driven CFO Strategies
OneStream introduced the Forward Finance Blueprint, a vendor‑driven framework that equips CFOs with AI‑enabled budgeting, forecasting and finance modernization tools. The playbook, built on five core dimensions, aims to turn AI potential into measurable business outcomes for its 1,800‑plus customers,...
Nvidia's Earnings Reveal AI Spending Shifts Beyond GPUs to Data Infrastructure
Nvidia reported record first‑quarter fiscal 2027 revenue of $81.6 billion, with data‑center sales climbing 92% to $75.2 billion. The company also announced a new reporting split that isolates a fast‑growing “ACIE” segment and highlighted networking revenue jumping 199% to $14.8 billion, underscoring a...
Pattern Launches Pi AI Engine to Automate Marketing and Operations for E‑commerce Brands
Pattern introduced Pi, an AI‑driven execution engine that automates marketplace tasks for DTC brands across Amazon, Walmart, TikTok Shop and eBay. Built on a proprietary 77 trillion‑point dataset, Pi already logged hundreds of thousands of automated actions, promising faster, data‑rich decision‑making.
Google Rolls Out Gemini‑Powered Ads in AI Search Results
Google introduced two Gemini‑driven ad formats—Conversational Discovery and Highlighted Answers—into AI Mode search results, letting advertisers serve AI‑generated explanations and chat‑based leads. The move expands Google’s ad inventory as search becomes more conversational.
Google Cloud and Amdocs Team Up to Extend Gemini AI to Telecom Enterprises
Google Cloud announced a partnership with Amdocs that places the latter's Telco Agents for Customer Experience on the Gemini Enterprise Agent Marketplace. The move targets telecom and media operators, offering AI‑driven workflow automation and new B2B revenue streams.
Goldman Sachs Report: Optical Networking Is the Next Mega Trend in AI Infrastructure
Goldman Sachs projects the optical networking market will swell to $154 billion by 2028, driven by exploding data‑center bandwidth needs of hyperscale AI workloads. The firm splits the opportunity, with scale‑up AI clusters representing $106 bn (69%) and co‑packaged optics (CPO) contributing...
Procore Launches AI‑Powered Datagrid Suite to Automate Construction Workflows
Procore Technologies rolled out an expanded Procore AI experience that embeds Datagrid intelligence, delivering agentic AI agents that can execute tasks, coordinate workflows, and act across construction projects in real time. The private‑beta launch aims to cut manual effort and...
Nigeria Deploys Meta‑Backed GovGuide AI Chatbot to Streamline Public Services
Nigeria’s Federal Ministry of Communications, Innovation and Digital Economy, together with Meta, NCAIR and Publica AI, launched GovGuide, an AI chatbot that uses Meta’s open‑source Llama models to deliver multilingual voice and text access to government information. The rollout is...
Workday Extends Sana AI to IT Service Management, Linking HR, Finance and IT
Workday announced that its generative AI assistant, Sana, will now handle IT service management (ITSM) requests, adding to its existing HR and finance capabilities. The move lets employees submit tickets, track incidents and receive automated guidance through a single conversational...
China Deploys $226 Million Offshore Underwater AI Data Centre Near Shanghai
China has begun commercial operations at the world’s first offshore wind‑powered underwater AI data centre off Shanghai, a $226 million project that houses nearly 2,000 servers and boasts a Power Usage Effectiveness of 1.15. The facility blends renewable energy with seawater...
JPMorgan Deploys AI Tools Across Global Investment Banking Division
JPMorgan Chase is rolling out artificial‑intelligence tools across its global investment‑banking business, said Paul Uren, the bank’s Asia‑Pacific head of investment banking. The move includes hiring more AI specialists, cutting traditional banker hires and leveraging Anthropic’s Mythos model under Project...
Harbor Capital Files Five AI‑Focused “Lab” ETFs Targeting Anthropic, DeepMind, Meta, OpenAI and xAI
Harbor Capital has filed with the SEC for five actively managed “Lab” ETFs that each target the ecosystem of a leading generative‑AI lab—Anthropic, Google DeepMind, Meta, OpenAI and xAI. The move expands the firm’s Harbor Scientific Alpha franchise and puts...
Tesla Rolls Out FSD 14.3.3 with Smoother Driving but Fewer Safety Alerts
Tesla began deploying Full Self-Driving software version 14.3.3, delivering smoother acceleration, more natural lane changes and a faster Smart Summon. At the same time, the update reduces the frequency of driver‑attention warnings, prompting criticism from safety advocates who fear the...
Utah Unveils Early Results From Statewide AI Prescription Renewal Pilot
Utah has released early results from a state‑wide experiment that uses artificial intelligence to automatically renew prescriptions for patients with chronic conditions. The pilot, closely watched by industry and regulators, marks one of the first large‑scale AI prescribing initiatives in...
DeepSeek Targets $10 Bn Valuation in 70 Bn Yuan AGI Funding Round
DeepSeek, the Hangzhou‑based AI lab, announced a 70 bn yuan ($10 bn) valuation target as it seeks at least $300 m in new capital. The round, the first to bring outside money into the lab, pits its open‑source, research‑first agenda against mounting pressure...
Semantic Layer Summit 2026 Positions Business Context as Core Infrastructure for Enterprise AI
More than 6,000 data leaders gathered at the Semantic Layer Summit 2026, where industry giants like Snowflake, Databricks and Anthropic affirmed the semantic layer as the foundation for accurate, governed enterprise AI. The summit underscored the need for shared business...
Microsoft and EY Pledge $1 B to Accelerate Enterprise Adoption of Agentic AI
Microsoft and EY have announced a $1 billion joint investment aimed at speeding enterprise adoption of agentic AI. The partnership will combine Microsoft’s cloud and AI platforms with EY’s consulting reach to create go‑to‑market programs, services and training for corporate customers,...
What Happens When Your AI Agent Interacts With Everything
MCP’s standardized protocol lets developers attach dozens of tools to an AI agent with minimal code, but model reasoning does not scale accordingly. A May 2026 LongFuncEval benchmark showed tool‑count, response length, and conversation depth can slash performance by up to...
Will We Really Put Data Centers in Space?
Major tech firms are eyeing orbital data centers (ODCs) as a way to sidestep terrestrial power bottlenecks and regulatory delays. The economic case hinges on SpaceX’s Starship achieving launch costs near $50 per kilogram, which would make space‑based solar power...
AI Is Being Used to Resurrect the Voices of Dead Pilots
The National Transportation Safety Board (NTSB) temporarily shut down its public docket after AI tools recreated the voices of pilots killed in UPS Flight 2976 using a spectrogram of the cockpit voice recorder. YouTuber Scott Manley demonstrated that the image...
XAI $15 Billion Per Year AI Rental to Anthropic Details
Elon Musk’s xAI is leasing a substantial portion of its Colossus 2 super‑computer to Anthropic, creating a $15 billion annual revenue stream. The deal covers roughly one‑third of the 550,000‑GPU, 1.5‑GW facility, with rental rates around $6 per hour. At full utilization...
Gemini Gained More Users in Q1 than Perplexity, Claude, Meta AI, DeepSeek, and Grok Combined as ChatGPT Referral Share Contracted
BrightEdge data shows Google Gemini captured more AI referral traffic in Q1 2026 than the combined totals of Perplexity, Claude, Meta AI, DeepSeek and Grok. ChatGPT’s share of AI referrals fell for the first time, dropping from 89.2% to 81.4%,...
YouTube Launches AI-Powered “Ask YouTube” Conversational Search and Adds Gemini Omni Remixing to Shorts
YouTube unveiled “Ask YouTube,” an AI‑powered conversational search that aggregates long‑form videos and Shorts into structured answers, initially rolling out to U.S. Premium members aged 18 and older. At the same event, Google integrated Gemini Omni into Shorts Remix and...
Dataiku Launches Cobuild on Snowflake for Governed AI Workflows
Dataiku has introduced Cobuild on Snowflake, a platform that translates natural‑language business requests into visual AI workflows, agents and applications running inside Snowflake. The offering leverages Snowflake Cortex’s large‑language‑model access and Dataiku’s orchestration layer to embed lineage, versioning and approval...
Your AI Agents Need a Terminal, Not Just a Vector Database
Researchers introduced Direct Corpus Interaction (DCI), a technique that lets AI agents query raw files via terminal commands instead of relying on vector embeddings. By bypassing semantic retrievers, DCI enables exact string, number, and pattern searches, improving multi‑step task performance....
18 Ways AI Is Rewriting the Future of Financial Services: Key Insights From the AWS Financial Services Symposium
The AWS Financial Services Symposium showcased 18 AI-driven transformations reshaping banking, insurance and capital markets. By 2025, 45% of IT leaders rank generative AI as their top budget priority, and nearly all firms plan to grow AI spend through 2026....
Free 435‑Lesson AI Engineering Curriculum From Math to Production
this guy literally put a full AI engineering curriculum on GitHub and made it 100% FREE 🤯 435 lessons. 20 phases. 320 hours. The rule that makes this curriculum completely different: Every algorithm gets implemented from raw math before a single framework gets imported. You build...
NineHertz Names Rohit Rao CTO to Accelerate AI‑Native Platform Strategy
NineHertz announced Rohit Rao as its new Chief Technology Officer, tasking him with deepening AI integration, scaling agentic capabilities, and sharpening platform engineering. The hire underscores the firm’s push toward AI‑native delivery for global enterprises.
Clariti AI Studio Launched to Speed Municipal Housing Approvals
Clariti announced the launch of Clariti AI Studio, a hands‑on program that trains municipal staff to embed artificial intelligence into permitting workflows. The initiative follows a pilot in Honolulu that slashed residential plan‑review times by 70%, and it aims to...
IManage Launches AI‑Ready Knowledge Platform at ConnectLive 2026
iManage announced its next‑generation AI‑ready knowledge‑work platform at ConnectLive 2026, highlighting AI‑specific controls, a new context fabric, and a surge to 78% cloud adoption across its global base. The rollout follows a year that added 90 new customer logos and...
Spotify and Universal Music Group Launch AI-Powered Cover and Remix Tool
Spotify and Universal Music Group (UMG) unveiled a generative‑AI partnership that lets Premium users produce licensed covers and remixes of catalog songs. The deal, announced at Spotify’s Investor Day, includes revenue‑sharing for artists and aims to boost discovery and engagement...
AI Implementation Is a Marathon, Not a Sprint
Seoul National University Bundang Hospital CIO Dr. Seyoung Jung cautioned healthcare leaders that AI adoption is a marathon, not a sprint. Executives must first pinpoint precise clinical tasks where AI can add value before committing to expensive tools. He emphasized...
Why I Use Gemini 3.0 Instead of ChatGPT for Multi-Step Agents (And How to Route Work to the Right AI)
The author switched from ChatGPT to Gemini 3.0 for building multi‑step AI agents, noting faster execution and cleaner context handoffs, especially when working with Google Workspace tools. Multi‑step agents chain actions such as Gmail search, calendar lookup, web research, and...