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:
Anthropic Releases Mythos-Like Model Without Cyber Capabilities
Anthropic announced the public rollout of Fable 5, a version of its Mythos model that blocks cybersecurity and biology queries through built‑in guardrails. For restricted topics, the Claude chatbot will defer to the Opus 4.8 model. Simultaneously, Anthropic released Mythos 5, a cyber‑capable variant available only to roughly 200 vetted organizations via Project Glasswing. The move comes as Anthropic prepares for an IPO and seeks to balance safety concerns with market demand for advanced AI capabilities.
Anthropic Releases Its First Mythos-Class Model Claude Fable
Anthropic unveiled Claude Fable 5, the first widely released model from its Mythos class, touting superior performance on software engineering, knowledge work, and vision tasks, especially as complexity grows. The launch is enabled by new safeguards that block high‑risk queries in...
Anthropic Releases a Version of Its Vaunted Mythos Model to Developers
Anthropic unveiled Claude Fable 5, a guard‑railed public version of its Mythos model, claiming it surpasses all previously released models in software engineering, knowledge work, and vision tasks. The model broke 90% on Hex’s core benchmark for complex, long‑running analytical tasks...
Anthropic’s New Model Is Mythos on a Leash
Anthropic announced Claude Fable 5, a public version of its unreleased Mythos model wrapped with new safety guardrails. The model redirects high‑risk topics such as cybersecurity and biology to the older Claude Opus 4.8, limiting its most dangerous capabilities. Anthropic says internal...
Claude Fable 5 Available Today in Microsoft Foundry: Powering the Next Era of Autonomous Agents
Anthropic’s Claude Fable 5 is now available through Microsoft Foundry, extending Mythos‑level AI to all Azure customers. The model is built for long‑running, multi‑stage tasks such as complex code refactoring, deep research synthesis, and document‑heavy workflows. Integrated with GitHub Copilot,...
Anthropic's New Claude Fable 5 Is a Nerfed Mythos with Guardrails Attached
Anthropic unveiled Claude Fable 5, a safety‑filtered version of its high‑performance Mythos model, making Mythos‑class capabilities available to general users. The new model automatically falls back to Opus 4.8 when prompts enter restricted cybersecurity or biology domains. Pricing is set at $10 per...
Anthropic’s Claude Fable 5 Is a Version of Mythos the Public Can Access Today
Anthropic has launched Claude Fable 5, the first publicly accessible version of its Mythos model, featuring built‑in safety limits that block high‑risk topics and fall back to Claude Opus 4.8. The rollout begins as a free preview for Pro, Max, Team, and...
Apollo, Blackstone Fund AI Boom
Apollo Global Management and Blackstone have teamed up to create a special‑purpose vehicle that will raise roughly $35 billion of private‑credit to finance the purchase of custom AI chips. The capital will initially be used by Anthropic, which will lease Broadcom‑Google‑built...
AI Uncovers Top Earbuds' Biggest Consumer Complaints
I gave an AI agent one task: read every 1-2 star review on Amazon's 5 best-selling earbuds and tell me why people hate them. It came back with the top 20 complaint phrases, grouped into 5 product themes. One run. It's called...
Facewatch Wants to Bring Live Facial Recognition to UK Pharmacies
Facewatch, a UK facial‑recognition surveillance vendor, is launching PharmacyProtect, a live facial‑recognition system tailored for community pharmacies. The solution, to debut at the Pharmacy Show 2026 in Birmingham, aims to curb theft of controlled medicines and reduce aggression toward staff,...
Apple WWDC 2026 Recap: What’s Coming to iPhone, Mac and Other Devices
Apple’s WWDC 2026 keynote unveiled a consumer‑focused artificial‑intelligence strategy, centering on a revamped Siri assistant. The company announced on‑device AI chips that accelerate Siri’s responsiveness and introduced a suite of AI‑driven features across iOS, macOS and other platforms. The event, lasting...
ADWEEK, Code and Theory, and ElevenLabs Are Bringing You an AI Radio Station at Cannes
ADWEEK, creative agency Code and Theory, and voice‑AI firm ElevenLabs have launched ai.rwaves, an AI‑powered radio station covering Cannes Lions from June 21‑26. The stream, built on Stagwell’s AI operating system called The Machine, delivers curated news, gossip, award tallies...
How Scientists Are Using AI to Analyze the Universe
Scientists at the Harvard‑Smithsonian Center for Astrophysics have launched the AstroAI Institute to apply artificial intelligence to massive astronomical datasets. The institute’s custom AI models can cluster data and surface unexpected patterns without any pre‑written code, dramatically speeding the hunt...
Mosaic Clinical Technologies™ Launches Mosaic Reporting™, an AI-Native Reporting Platform Built for Radiology Workflows
Mosaic Clinical Technologies, a Radiology Partners subsidiary, has commercially launched Mosaic Reporting, an AI‑native platform that builds radiology reports in real time as images are interpreted. Powered by Cognita Imaging’s in‑house foundation models and embedded in the MosaicOS cloud‑native operating...
This AI Agent Startup Ditched Anthropic for DeepSeek — and Says It’s Saving Millions
Lindy, a no‑code AI agent platform, announced it has migrated its entire production workload from Anthropic’s models to DeepSeek v4. The switch, after a 6‑9‑month evaluation, is projected to save the company millions of dollars while delivering better performance on core...
Meta Is Launching a Trades Academy to Bolster AI Infrastructure Build-Outs
Meta announced a $115 million America’s Workforce Academy to train workers for AI data‑center construction. The free program, piloted in Louisiana, Ohio, Indiana and Texas, promises participants a job after completion, though details on employers and union status remain unclear. The...
Korn Ferry Highlights AI 'Beach‑Buddy' Agents Covering Work While Employees Vacation
Korn Ferry says companies are rolling out AI "Beach‑Buddy" agents to manage routine correspondence, scheduling and document synthesis while employees are on holiday. The move aims to blunt the typical productivity dip and ease post‑vacation catch‑up, but it also raises...
Zillow Launches AI‑Powered Marketing on Google NotebookLM to Reach Home Buyers
Zillow has introduced a branded research assistant on Google’s NotebookLM, turning the generative‑AI tool into a marketing channel that offers personalized answers and podcast‑style audio overviews for prospective home buyers. The move signals a shift toward AI‑generated content at scale...
NTT, SK Group and Partners Launch $440 Million AI Optical Network Fund
Japanese telecom giant NTT, South Korea’s SK Group, Taiwan’s Chunghwa Telecom and the Development Bank of Japan announced a $440 million IOWN AI Fund. The vehicle will back startups developing photonics‑electronics convergence, AI semiconductors and advanced AI models that run on...
Compute‑Normalized Benchmarks Needed to Truly Gauge Model Progress
Great post. So much about model performance is a function of how much compute you’re doing at inference time. This means compute-normalized benchmarks is the only logical path forward. And yet, the challenge is it’s a lot harder than it...
Reco Appoints COO Zoe Hillenmeyer as It Expands AI Agent Security Footprint in Texas and UKI
Reco announced Zoe Hillenmeyer as its new chief operating officer, added senior sales leaders, and opened a Texas office while planting a team in the United Kingdom and Ireland. The moves aim to scale go‑to‑market execution as more Fortune 500...
NVIDIA Teams with LG to Build AI‑powered Humanoid Robots in South Korea
NVIDIA and LG Group unveiled a partnership to create an AI factory and develop AI‑powered humanoid robots, including LG's CLOiD home robot, using NVIDIA's Isaac simulation and GR00T foundation models. The deal, announced on June 7, ties NVIDIA’s full‑stack AI...
Apple Shakes Hands with Google for Powering Siri
Apple announced a partnership with Google to embed Google’s Gemini large‑language model into Siri, creating a standalone AI assistant app for iPhone, iPad and Mac. The integration will run alongside Apple’s own foundation models, delivering multimodal capabilities that handle voice,...
Tools for Humanity Cuts Staff as Revenue Stalls, Despite $2.5B Valuation
Tools for Humanity, the biometric ID venture co‑founded by OpenAI CEO Sam Altman, is laying off employees after failing to turn its iris‑scanning Orb into a sustainable revenue stream. The move comes despite a $2.5 billion valuation and backing from Andreessen...
Apple's Safari 'Notify Me' Alerts Aim to Boost E‑commerce Conversions
At WWDC 2026 Apple unveiled a new Safari feature called Notify Me that monitors product pages for stock or price changes and pushes real‑time alerts to users. The AI‑powered tool is positioned to increase conversion rates for online retailers while keeping...
FINNY Launches Enterprise AI Growth Platform with Mercer Advisors
FINNY, the New York‑based AI prospecting startup, unveiled an enterprise growth platform aimed at centralized marketing and business‑development teams of large wealth‑management firms. Mercer Global Advisors, which oversees roughly $83 billion in assets, became the flagship user, scaling the tool from...
Karias Health Launches Faith, an AI Care Companion Built for Real-Time Healthcare Decisions
Karias Health launched Faith, an AI care companion that listens to clinical visits, summarizes them in plain language, and guides members to high‑quality, low‑cost next steps. Faith integrates with the K‑Card platform to provide real‑time decision support before, during, and...
UK Doctors Could Face Claims Over NHS AI Errors
UK doctors and the NHS face potential negligence lawsuits if AI tools make diagnostic or treatment errors, prompting the Medical Protection Society to urge that AI be classified as a product under the Consumer Protection Act 1987. The government and...
4 Ways Researchers Are Collaborating with Co-Scientist to Solve Big Problems
Researchers are adopting Co‑Scientist, a multi‑agent AI platform that generates, debates, and refines scientific hypotheses across life‑science challenges. The system orchestrates specialized agents through idea generation, peer‑review style debate, and evolutionary refinement, all coordinated by a supervisor agent. Early collaborations...
NHS Prescribes Half a Million Copilot Licenses For Its Paperwork Headache
NHS England will equip roughly 505,000 clinicians and support staff with Microsoft Copilot after a pilot showed an average saving of 43 minutes per day on administrative work. Licenses will be allocated per trust, typically starting with about 2,000 seats,...
See What 3 Builders Are Making with Gemma 4
Google’s Gemma 4, an open‑source large language model, has surpassed 150 million downloads, showcasing rapid adoption among developers. The release adds Multi‑Token Prediction, a 12 B unified model, and Quantization‑Aware‑Training checkpoints, all under an Apache 2.0 license for unrestricted fine‑tuning. Builders are already leveraging...
The New Cost of AI: Why CFOs Are Becoming AI Budget Gatekeepers
The article explains how CFOs are evolving into the primary gatekeepers of AI spending as enterprises move from experimental pilots to operational deployments. While early AI projects were funded through innovation budgets, the shift to ongoing subscription fees, integration services,...
Supercharging Data Foundations for Agentic AI with Fivetran and Redis
The DBTA roundtable highlighted that enterprise AI agents now depend on robust data foundations rather than just model accuracy. Fivetran and Redis outlined key architectural components—semantic layers, knowledge graphs, vector retrieval, modern data platforms, and real‑time pipelines—to supply fresh, governed...
Asia’s Largest Outsourcer to Slow Hiring as AI Reshapes Industry
Tata Consultancy Services, Asia’s largest IT outsourcer, announced it will deliberately curb new hires as it scales artificial‑intelligence agents to match its roughly 500,000‑strong workforce. Chairman Natarajan Chandrasekaran said the goal is to have an equal number of AI workers...
TPC26: Toward Scientific AI Platforms at HPC Facilities
At the TPC26 conference, leaders from national labs, supercomputing centers and industry discussed building scientific AI platforms that deliver inference services and AI agents on HPC facilities. The National Research Platform showcased a token‑based model offering nine open AI models,...
Centrical Raises $39M to Transform Frontline Performance for the AI Era Workforce
Centrical announced a $39 million Series D round led by Leeds Illuminate and Kingfisher Investment, with existing backer JVP participating. The funding will accelerate global expansion and the rollout of AI‑driven tools that automate coaching, role‑play simulations, and performance interventions for both...
I Tested iOS 27's New AI Photo Editing Tools as a Skeptic - and the Results Surprised Me
Apple unveiled three AI‑powered photo‑editing tools—Clean Up, Extend, and Reframe—in the iOS 27 developer beta released after WWDC 2026. Clean Up now removes unwanted objects with higher accuracy, while Extend adds AI‑generated space around a subject and Reframe shifts the perspective as if the...
Cisco Sees Rising Role for Service Providers in AI's Inferencing Era
Cisco warned that AI inference traffic could account for roughly 25% of total network traffic by 2035, shifting demand from power‑hungry data centers to low‑latency edge environments. The company sees cable operators and telcos as prime candidates to host “inferencing...
Stay Ahead: Master Emerging AI Trends for Business
🚀 What’s next in AI? AI is evolving at an incredible pace, and the organisations that stay ahead of the curve will be the ones best positioned to innovate, compete and grow. I'm delighted to be partnering with LinkedIn Learning...
Claude Mythos Turns N-Days Into N-Hours With Rapid Exploit Creation
Anthropic announced that its Claude Mythos Preview model can automatically generate working exploits for known vulnerabilities in minutes to hours. In tests, Mythos produced 16 functional exploits for Firefox and Windows, including eight Windows privilege‑escalation exploits within 18 hours. The...
Al-Designed CTV Saves 100,000 Liters of Fuel, 258 Tons of CO2 a Year
Compute Maritime’s UK‑funded GenDSOM project used its NeuralShipper AI and additive‑manufacturing tools to create a 32.5‑meter twin‑hull crew transfer vessel for offshore wind. The AI‑optimised hull and diesel‑electric hybrid system deliver an 11.1% cut in annual fuel use—about 101,671 liters—and an...
AI Coding Adoption Hits 97% but Governance Lags Behind
A recent UserEvidence survey of 831 engineers shows 97% of software teams now rely on AI coding assistants such as GitHub Copilot and Claude Code, yet only about 30% have formal governance policies. While 92% of respondents credit the tools with...
Fluid, Natural Voice Translation with Gemini 3.5 Live Translate
Google unveiled Gemini 3.5 Live Translate, an audio‑first model that delivers near‑real‑time speech‑to‑speech translation in more than 70 languages. The system streams translated audio continuously, preserving intonation and staying only a few seconds behind the speaker. It launches today via the Gemini Live...
China Sets Out $295B AI Investment Plan
China is allocating roughly CNY2 trillion (about $295 billion) over the next five years to construct a nationwide network of AI‑focused data centres. State‑owned telecom operators China Mobile and China Telecom will manage the sites, while Huawei is expected to provide at...
AI Flags Breast Cancer Risk up to Ten Years Early
AI of radiologist read as normal mammograms shown to detect high-risk of breast cancer out several years "Artificial intelligence scores from sequential mammograms in individuals diagnosed with breast cancer showed elevated scores up to 10 years before diagnosis" a retrospective study, but supported...
Anthropic Launches Mythos Upgrade for Partners, Safe Version for All
Anthropic Offers Mythos Upgrade for Cyber Partners and a ‘Safe’ Version for the Rest of You https://t.co/eX1Pzy8kFe https://t.co/Q79b34Jdz4
UK Government Launches AI Growth Lab with Legal as Its First Focus Area
On 9 June the UK government unveiled the advisory AI Growth Lab, a new platform that will bring together regulators to provide clear guidance on how existing rules apply to emerging AI uses. Its inaugural focus is LawTech, targeting legal and...
Apple's New AI Explained: Gemini Distillation, EU Troubles and Third-Party Clouds
Apple announced a revamped Apple Intelligence suite built on five new foundational AI models co‑developed with Google. The partnership taps Google’s Gemini technology and runs part of the workload on Google Cloud, promising more personalized experiences across iOS, macOS, and...
Anthropic Says Mythos Is Too Dangerous for the Public. It Just Gave 150 More Organisations Access.
Anthropic warns that its Mythos AI model is too dangerous for public release, yet it has broadened access to roughly 200 organizations across 15 countries, adding 150 new participants in June. Mythos can uncover thousands of zero‑day vulnerabilities, chain them...
Reimagining Client Experience Through AI-Powered Product Innovation
Forrester has rebranded its 2023 generative‑AI offering as Forrester AI and embedded it directly into Microsoft Teams and Microsoft 365 Copilot. The platform now features chain‑of‑thought reasoning, support for over 200 languages, and integrated research assets that let clients interact with...