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
Snowflake and Collibra Expand Integration to Bring Governed Business Context and Semantics Across the Snowflake AI Data Cloud
Snowflake and Collibra announced an expanded, bi‑directional integration that embeds Collibra’s governed business context and semantic models directly into Snowflake’s AI Data Cloud. The partnership synchronizes metadata, lineage, and access policies between Collibra’s AI Command Center and Snowflake Horizon Catalog, Cortex Analyst, and Cortex Agents. This enables natural‑language AI queries and agentic workflows to operate on trusted, up‑to‑date definitions, improving accuracy and compliance. The integration is in preview for select customers, with broader rollout planned for Q3 2026.
DeepSeek Lines up Its First Outside Money: A $7bn Round at up to $59bn
DeepSeek, the Beijing‑based AI lab that rose to prominence after its V3 and R1 models impressed Silicon Valley, is about to close its first external financing round. Sources say the company will raise roughly 50 billion yuan (~$7 bn), valuing it between...
AI and Advertising: Protecting Creativity in an Automated World
In a recent Adspeak episode, Adweek’s Kim Einan and Voya Financial’s Jim Cowsert discuss how AI can amplify advertising creativity without eroding brand identity. They argue that AI should serve as a data‑driven insight engine, enabling consistent, personalized campaigns while preserving the...
Telecom Is the Cheapest Way to Distribute AI – VEON
VEON, the Dubai‑based telecom operating in Bangladesh, Kazakhstan, Pakistan, Ukraine and Uzbekistan, is positioning its network as the cheapest distribution channel for AI‑enabled digital services. Digital offerings now account for over a quarter of group revenue and grew 57.7% YoY...
Snowflake Summit 2026- How Snowflake Is Making a Strategic Shift Towards Agentic AI
Snowflake announced a strategic pivot at its 2026 Summit, positioning the company as the control plane for agentic AI. CEO Sridhar Ramaswamy introduced an "agentic control plane" that orchestrates AI models, services, and enterprise data to work together seamlessly. The...
Machine Unlearning: The Technical Solution to AI Copyright Scandals
Machine unlearning technology lets AI teams erase the influence of specific training data without rebuilding models from scratch. The approach addresses privacy requests, copyright claims, and licensing errors by updating weights, retrieval layers, or safety filters. It offers a cost‑effective...
Bold Symposium At Stanford Illuminates The Future Of AI For Mental Health
Stanford’s AI4MH symposium on June 1, 2026 gathered researchers, clinicians, industry leaders, and policymakers to map the future of AI in mental health. Sessions highlighted the surge in consumer use of large‑language models—24‑33% of U.S. adults now turn to AI for mental‑health...
Mukesh Ambani’s JioStar Bets Big on AI Content with New Series Lineup
Mukesh Ambani’s JioStar, a Reliance‑Disney joint venture, is rolling out a slate of series created entirely by artificial intelligence. The move follows a successful AI‑generated retelling of the Mahabharat epic that attracted 6.5 million views on its first day, prompting the...
Inside Loxo: Driverless Delivery Vans in Europe
In this episode of Autonomy Insiders, host Daniel interviews Amin Amini, co‑founder and CEO of Swiss autonomous‑vehicle startup Loxo, about their Level‑4 driverless delivery vans focused on urban logistics. Amini explains Loxo’s strategy of targeting the massive European light‑commercial‑vehicle market...
Retailers Face Surge in AI-Generated Fake Damage Claims and Refund Scams
Retailers are confronting a surge in AI‑generated refund fraud, with generative tools enabling scammers to fabricate damage photos and false return narratives. Forter reports that these fake damage claims are now the fastest‑growing form of return abuse, amplified by flexible...
Q&A: Som Sinha, CEO, SAI
SAI, led by CEO Som Sinha, uses a Visual Language Model (VLM) to convert existing CCTV into a real‑time store‑intelligence platform for UK retailers. The solution delivers loss‑prevention alerts, shopper‑flow analytics and operational recommendations across roughly 1,000 stores, generating multi‑million‑pound...
Microsoft's Approach to LLM: MAI-Thinking-1
Microsoft’s AI team released a technical report on MAI‑Thinking‑1, a reasoning model that achieved 52.8% on SWE‑Bench Pro and 97% on AIME 2025, matching frontier‑size models. The model was trained on 30 trillion human‑written tokens without any external model distillation, emphasizing learning...
Snowflake CoCo Redefines Enterprise AI Development as the Coding Agent Built for Faster, Easier, and More Powerful Innovation Anywhere
Snowflake unveiled 26 new capabilities for its CoCo AI coding agent and introduced Snowflake Datastream, a fully managed Kafka‑compatible streaming service. CoCo now ships as a native desktop app, mobile app, Slackbot, and extensions for VS Code, Microsoft Excel and Claude...
Anthropic Expanding Access to Mythos AI Model to India Amid Rising Cybersecurity Demand: Report
Anthropic is extending its cybersecurity‑focused AI model Mythos to more than 15 countries, including India, Canada, and Japan. The rollout follows the controlled Project Glasswing pilot that gave roughly 150 organizations early access and uncovered over 10,000 high‑severity software vulnerabilities....
Seven Essential Guardrails for Safe AI Agent Deployment
The 7 AI Agent Guardrails Every Business Needs Before Things Go Wrong #AI #agents promise to #automate work, make decisions and transform business operations, but giving machines more #autonomy also creates new risks. From access controls and #audit trails to...
RACK: The Building Block Of The AI Infrastructure Boom
The RACK fund is positioning itself as a pure‑play vehicle for the AI infrastructure surge, targeting companies that supply chips, power, cooling, networking and real‑estate for data‑centers. Analysts expect data‑center capacity to triple by 2030, with roughly 70% of that...
Megaport Secures Four New AI Infra Contracts, to Raise $594 Million
Australian connectivity provider Megaport announced four new AI‑infrastructure contracts valued at A$458.9 million ($329 million) and launched an entitlement offer to raise A$827.3 million ($594 million). The contracts, slated to begin in the first half of 2027, will fund roughly A$369.5 million in capital spending...
Meta’s Cloud Computing Signal: How AI Data Center Spending Could Become A Hyperscaler Business
Meta plans to spend $125‑$145 billion on AI‑focused data centers in 2026, a scale comparable to hyperscalers. CEO Mark Zuckerberg told shareholders that any excess compute capacity could be sold as a cloud service, turning a potential cost center into a...
AI and the New Value Equation - the Three Phases of the AI Transformation Journey (2/3)
The article outlines three distinct AI‑generated value streams—subtractive cost cuts, additive revenue creation, and transformative industry‑changing impact—and ties them to an organization’s stage in its AI transformation journey. Most firms are currently harvesting subtractive value but struggle to move beyond...
2026 VIEW FROM THE TOP: Franz Inc.
Franz Inc. announced that AllegroGraph 9 now includes GraphTalker, a capability that transforms its neuro‑symbolic knowledge‑graph platform into a Context Graph for agentic AI. GraphTalker lets AI agents navigate RDF graphs, embeddings, and LLM outputs through a reasoning loop that...
Siemens and Partners Develop NVIDIA AI Data Center Reference Architecture
Siemens, NVIDIA, Fluence and nVent have released a reference architecture built around NVIDIA’s DSX Vera Rubin NVL72 platform to guide the construction of high‑density AI data centers. The design can handle up to 136 MW of total facility power with a...
Employees ‘Stay Silent’ over Harmful AI Errors at Work
A new Writer survey of 2,400 UK employees and executives reveals that 28% of staff have witnessed AI tools generate dangerously wrong, unethical or biased outputs at work. Yet three‑in‑ten employees feel unsafe reporting these issues, fearing retaliation, while 90%...
Will AI Ruin the Social Sciences — or Revolutionize Them?
Researchers across the social sciences warn that large language models are infiltrating survey research, with estimates that up to 45% of responses may be AI‑generated. The contamination threatens data validity and could undermine trust in behavioural studies. Journals are already...
Google Search Console AI Performance Reports and Controls to Block Your Content in AI Responses
Google Search Console is introducing a Generative AI performance report that tracks impressions, pages, countries, devices and timeframes for content appearing in AI‑driven search features. The report, currently limited to a subset of UK site owners, does not include click‑through...
Big Tech's AI Ambitions Pose a Major Power Test for Europe
SoftBank announced a €75 billion (~$81 billion) investment to build 3.1 GW of AI data‑centre capacity in France’s Hauts‑de‑France region by 2031, cementing the country’s role as Europe’s AI hub. France’s grid, powered over 60% by nuclear, can absorb the energy‑intensive workloads that...
Trends From the Trenches: AI-Ready Life Sciences Data
Life sciences firms are racing to adopt AI, but fragmented, unstructured data across instruments, on‑prem NAS, and multiple clouds creates silos that slow research. Hammerspace CTO Adam Marko explains that a unified, rule‑based data orchestration layer can present a single...
SymphonyAI Brings AI-Powered Assortment and Space Platform to Global CPGs, Compressing Category Review Cycles From Weeks to Days
SymphonyAI launched CINDE Assortment and Space, an AI‑driven platform that links assortment strategy, planogram execution, and in‑store compliance for consumer packaged goods firms. The solution compresses the traditional four‑to‑six‑week category review cycle into days by using computer‑vision photo‑to‑planogram loops and...
Infobip Releases White Paper to Guide Enterprise Voice AI Strategies
Infobip, the AI‑first cloud communications platform, released a new white paper titled “Future‑Proofing Your Investment in Voice AI: A Strategic Guide for Enterprises,” developed with analyst firm Omdia. The report highlights the rapid growth of the Voice AI market, projected...
‘The CGI Would Have Cost Millions. I Spent $2,000.’ Is Dreams of Violets AI Slop – or the Future of...
Iranian‑British filmmaker Ash Koosha is premiering *Dreams of Violets* at Tribeca, the first fully AI‑generated live‑action drama about the 2022 protest crackdown. Every visual element was created with generative AI, while Koosha wrote the script, composed the score and edited the...
Inside Alexandr Wang’s Bid to Revive Meta’s AI Edge
Meta has recruited Scale AI co‑founder Alexandr Wang to head its artificial‑intelligence division, tasking him with restoring the company’s lagging AI edge. Wang will oversee the scaling of large language models, improve data‑labeling pipelines, and drive cost‑efficiency after Meta’s $10 billion...
Phihong Group Unveils AI Power and Smart Energy Strategy Integrating ZEROVA Brand
At Computex 2026, Phihong Group announced its AI Power × Smart Energy strategy, merging its long‑standing power‑supply business with the ZEROVA smart‑energy brand. The rollout includes AI‑optimized power‑over‑Ethernet, modular power modules, redundant data‑center systems, and USB‑C high‑density battery chargers for robotics and...
YY Group Invests in NVIDIA Blackwell Infrastructure to Accelerate Workforce AI Strategy
YY Group Holding announced a strategic investment in NVIDIA Blackwell‑based infrastructure featuring GeForce RTX 5090 GPUs. The on‑premise hardware will power proprietary large‑language‑model fine‑tuning, ultra‑low‑latency candidate‑to‑job matching, and secure offline experimentation. By leveraging CUDA, QLoRA, and TensorRT, YY aims to build...
Uber's AI Bill Got so High that It Is Now Limiting Claude Code Usage to $1,500: Report
Uber has imposed a $1,500 monthly cap on employee usage of AI coding tools such as Anthropic's Claude Code and Cursor after exhausting its annual AI budget within four months. The limit applies per tool and is monitored via an internal...
AI Can Speed up Grievance Resolution in BFSI, but Trust Still Needs a Human Touch
Artificial intelligence is reshaping grievance redressal in India's BFSI sector, automating complaint categorisation, urgency assessment and routing to cut resolution times dramatically. The RBI reports a 13.55% rise in complaints to 1.33 million in FY25, with 91% filed through digital channels,...
VANDAL Partners with University of Technology Sydney to Explore Creative Generative AI in the Future of Hybrid Animation
VANDAR has teamed up with the University of Technology Sydney to embed generative AI into hybrid animation workflows. The partnership, funded by an Australian Research Council Discovery Project, will launch a series of school‑focused animated films that spotlight scientific figures....
Agentic AI Needs Infrastructure—Vora IQ Delivers
Will.i.am just started teaching "The Agentic Self" at Arizona State. The frame: you operate AI as a system, not query it as a tool. But an agentic self only matters if you have agentic infrastructure. Students need somewhere to actually run that...
Google Launches Free Offline AI Dictation, Rattling Startups
SPEECH-TO-TEXT STARTUPS ARE OFFICIALLY SWEATING RIGHT NOW. Google just dropped its own AI voice dictation app for macOS and iOS. You get all the premium feats of a paid speech-to-text app without any subscription. 100% FREE. Runs offline. Powered by Gemma 4. Download link...
Microsoft Embraces OpenClaw
At Microsoft Build, CEO Satya Nadella announced that the company will fully adopt the open‑source AI framework OpenClaw for its upcoming Scout agent, slated for a wider release this summer. Scout will run on OpenClaw and Microsoft will feed security guardrails...
10 Hacks Every Perplexity User Should Know
The LifeHacker piece outlines ten practical hacks that transform Perplexity from a simple Q&A tool into a robust research assistant. It highlights built‑in features such as the Perplexity Computer for background processing, the Comet AI browser for live web access,...
New MAI Models Enable Companies to Join Frontier AI
With the new MAI models and Frontier Tuning capabilities we announced today, we're focused on helping every company move from just consuming a frontier model to fully participating at the frontier. https://t.co/GJwGCypRfj
The Details of Trump’s Long-Awaited, Scaled-Back AI Order
President Donald Trump signed a pared‑down executive order on June 2, 2026 aimed at bolstering cybersecurity against AI‑driven threats. The order directs federal agencies to prioritize AI‑enhanced cyber defenses and creates a voluntary AI cybersecurity clearinghouse for sharing vulnerability patches....
IREN and BE Networks Use NVIDIA DSX Air to Simulate 50,000‑GPU AI Factory
BE Networks and IREN Limited have teamed up with NVIDIA to run a digital‑twin simulation of an AI cloud that will house more than 50,000 Blackwell Ultra GPUs. The DSX Air platform lets the partners validate network topologies, automation workflows...
"The Value Didn't Arrive": Bain Finds Cost-Savings From AI Are Falling Far Short Of Projections
A Bain Global Survey of 951 firms with revenues over $100 million reveals AI‑driven automation is delivering far less cost‑savings than expected. Only 40% of respondents reported reductions of 10% or less, and 44% are planning the next wave of AI...
Sema4.ai’s Autonomous Agent-Building Platform Gets Simpler to Use, Adds Deeper Business Context and More
Sema4.ai unveiled a major overhaul of its autonomous agent‑building platform, adding a no‑code Agent Builder, deeper Business Context Layer, and persistent memory. The update expands integrations through an MCP Access Gallery that connects to Snowflake, Slack, GitHub, Google Workspace and...
Anthropic Grants Project Glasswing Access to 150 More Companies, with a Focus on Critical Infrastructure
Anthropic announced that Project Glasswing will now include 150 additional companies, targeting critical‑infrastructure sectors such as power, water, healthcare, communications and hardware. The expansion aims to boost AI‑driven vulnerability hunting, which has already uncovered thousands of flaws across major tech...
AI Must Work for an Ageing Workforce, Not Against It
Australia’s workforce is aging, with participation increasingly extending beyond the traditional retirement age of 64. Ministers announced an AI Employment and Workplaces Forum to address how artificial intelligence can support, rather than hinder, older employees. The article argues that older...
GOODSTUPH Singapore Opens the Kitchen on AI Creativity with New Masterclass Series
GOODSTUPH Singapore has unveiled Open Kitchen, the city’s first rotating dine‑and‑learn AI marketing masterclass, powered by its AI‑first division AI.YAH. The inaugural session runs on June 9 from 6 pm to 10 pm at Overlap, The Working Capitol, with tickets priced at S$388...
Google Is Quietly Buying Code From Play Store Developers to Train AI
Google has quietly launched a pilot program that pays Android app developers for access to their codebases, including active and archived projects, to improve its AI coding tools. The offer, delivered via email, guarantees developers retain full IP rights under...
Amazon Rolls Out Smart Bid V2, Upending Sponsored Products PPC Strategies
Amazon has deployed its upgraded Sponsored Products bid automation engine, Smart Bid v2, across all U.S. Seller Central accounts. The AI‑driven system adjusts bids per auction using demand forecasts, driving CPC spikes of up to 14% while delivering ROAS gains...
OpenAI Expands Codex Beyond Coding with Role-Specific Plugins for Analysts, Marketers, Sales, and Other Knowledge Workers
OpenAI is extending its Codex platform beyond software development with six new role‑specific plugins aimed at analysts, marketers, salespeople and other knowledge workers. The plugins aggregate 62 third‑party applications and 110 AI‑driven skills covering data analytics, creative production, sales, product...