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:
Legora Acquires Swedish AI Legal Research Start-Up Qura
Legora has completed the acquisition of Swedish AI legal‑research start‑up Qura, adding its 10‑person team and proprietary semantic search engine to Legora’s platform. The deal follows Legora’s recent $550 million Series D round that valued the company at $5.5 billion and its earlier purchase of Canadian AI firm Walter AI. Qura’s technology enables deep, context‑aware searches across millions of documents and is already live in 27 jurisdictions, reporting a 40% month‑over‑month revenue increase. Legora plans to leverage the combined assets to accelerate U.S. market penetration and broaden its global legal‑AI stack.
Inside The Standard's Bengaluru Bet: Building the AI-Ready Workforce of Tomorrow
The Standard opened a global capability centre in Bengaluru to accelerate AI‑driven modernization of its 119‑year‑old insurance platform. The centre will house AI engineering, cloud, data analytics, and API integration teams that support the recent Allstate Employer Voluntary Benefits acquisition....
Refining Your Go-to-Market Plan with an AI GTM Agent
Enterprises are turning to agentic AI agents to embed go‑to‑market (GTM) functions—such as approval routing, content cleanup, and learning checks—directly into daily workflows. By unifying sales, marketing, enablement and RevOps around a single AI‑driven operating rhythm, companies can keep launches,...
AI Video Costs Drop to $0.10/Second, No Guesswork
The "blind generation" era of AI video is officially over. @TopviewAIhq just slashed Seedance 2.0 to $0.10/second... and completely changed the math for bulk creators. The real pain of AI video was never the prompt. It was the coin flip every time...
Sam Bock, Relativity: 3 Adversaries You Might Meet Negotiating an AI-Friendly ESI Protocol
Sam Bock’s Relativity piece warns that an AI‑friendly ESI protocol can make or break an e‑discovery project. He outlines three typical adversaries—opposing counsel, IT teams, and data custodians—who can complicate negotiations. The article stresses early collaboration between clients and technologists...
Image‑2‑Thinking Introduces Agentic Loop for Text‑to‑Image
btw in talking to friends the best framing for how to discuss GPT-Image-2-Thinking taking multiple tens of mins for generation and being able to oneshot QR codes and diagrams and logos and foods and faces.. ...is that Image-2 is a new...
Luminance and LexisNexis Unite AI Tools to Streamline Contract Lifecycle Management
Luminance and LexisNexis announced a strategic alliance that embeds LexisNexis Protégé and Lexis+ research capabilities inside Luminance’s contract workflow platform. The integration gives corporate legal departments instant access to a combined data set of over 200 billion legal documents, aiming to...
AMD and Intel Consumer CPU Prices Surge 10% as AI‑Driven Shortages Grip Market
AMD and Intel have raised consumer CPU prices by about 10% in the past month, reflecting tight supply and soaring AI‑driven demand. The surge, mirrored in server chips, signals a multi‑year pricing cycle that could reshape PC and data‑center cost...
Google Cloud Signs Multi‑Billion Nvidia GB300 Deal with Thinking Machines
Google Cloud announced a multi‑billion‑dollar agreement with Mira Murati’s 14‑month‑old Thinking Machines Lab to power its AI workloads on Nvidia’s GB300 chips. The deal, valued in the single‑digit billions, locks in A4X Max VMs that promise double the speed of...
Covered California Teams with Google and Deloitte to Cut Eligibility Work by 40%
Covered California announced an expanded partnership with Google Public Sector and Deloitte to embed Google Cloud's Document AI into its CalHEERS platform. The AI upgrade reduces manual processing by 40%, delivering real‑time eligibility verification for millions of Californians and strengthening...
Hanoi Targets 95% E‑ID Coverage by Year‑End in New Digital Transformation Plan
Hanoi has approved a sweeping digital transformation plan that will require all municipal agencies to adopt artificial intelligence from Oct. 1 and aims to issue electronic IDs to 95% of its residents by the end of 2026. The initiative also sets...
OpenAI Unveils ChatGPT Images 2.0 with Text, Slides and Multilingual Generation
OpenAI rolled out ChatGPT Images 2.0 today, extending its image model to render dense text, generate slide decks and support Asian-language scripts. The upgrade, available via API from day one, is aimed at educators, developers and EdTech builders seeking richer...
Indiana University Opens Free GenAI 101 Course to Global Learners
Indiana University’s Kelley School of Business has opened its flagship GenAI 101 certificate program to anyone worldwide at no cost. The eight‑module course, which has already enrolled more than 114,000 students, staff and faculty, is positioned as a public‑sector response...
Harness Teams with Google Cloud to Fuse AI Into Software Delivery Pipelines
Harness announced at Google Cloud Next that it is integrating Google Cloud's Developer Connect with its Software Delivery Knowledge Graph, giving joint customers a unified, AI‑ready view of the entire delivery lifecycle. The move, highlighted by Harness winning the 2026...
Meta Secures Up to 1 GW/100 GWh Ultra‑Long‑Duration Storage From Noon Energy for AI Data Centers
Meta Platforms has agreed to reserve up to 1 GW of ultra‑long‑duration energy storage – equivalent to 100 GWh – from Noon Energy. The partnership starts with a 25 MW/2.5 GWh pilot slated for completion by 2028 and is aimed at delivering multi‑day renewable...
Microsoft Invests $25B in AI to Embed Trust and Inclusivity Into the Australian Psyche
Microsoft announced a $25 billion, three‑year investment to expand AI infrastructure, cloud capacity and cybersecurity in Australia, emphasizing trust and inclusive growth. The plan includes adding 29 Azure data‑centre sites, extending the Cyber‑Shield program, and delivering AI skills to 3 million Australians...
Sullivan & Cromwell Apologises After AI Hallucinations Taint Court Filing
Sullivan & Cromwell issued a formal apology to Chief Judge Martin Glenn after an emergency bankruptcy motion contained about 40 AI‑generated citation errors. The firm acknowledged lapses in its AI review protocol, highlighting growing concerns over generative‑AI reliability in legal...
Sony AI’s Robot Ace Beats Elite Humans in Table Tennis, Marking First Real‑World AI Victory
Sony AI announced that its robot Ace defeated elite human opponents in table tennis, winning three of five matches under official ITTF rules. The achievement demonstrates a new level of real‑world AI performance, extending the company’s prior virtual‑domain successes into...
Nasdaq Hits Record 24,596 as AI and Big‑Tech Earnings Drive Rally
The Nasdaq Composite closed at a fresh all‑time high of 24,596.22, propelled by AI‑centric chip makers and big‑tech earnings. The rally was amplified by President Trump’s extension of the Iran ceasefire, which eased geopolitical tension and revived investor confidence.
Northrop Grumman Slashes Spacecraft Design Time to Hours with AI
Northrop Grumman announced that AI integration has reduced spacecraft design cycles from years to hours, a speedup of roughly 100 times. The breakthrough, achieved with Flexcompute and NVIDIA, promises faster, more precise space missions and could reset industry timelines.
Michael Gennaro: AI Is Outpacing the Systems Built to Govern It, Stanford Report Finds
Stanford’s 2026 AI Index reports AI adoption in nearly 90% of organizations and $286 billion in private investment, outpacing governance frameworks. The report shows U.S. software developer employment for ages 22‑25 dropped almost 20% since 2024, while older developers saw headcount...
Afresh Secures $34 Million to Scale AI Grocery Platform Across U.S. Stores
Afresh announced a $34 million funding round led by Just Climate and High Sage Ventures to accelerate deployment of its AI grocery platform across more than 12,500 departments in 40 U.S. states. The capital will fund broader retailer rollouts and the...
Whitepaper: A Roadmap for Responsible AI Adoption in Australia’s Public Service
The Australian Public Service (APS) has released a new whitepaper outlining a roadmap for responsible AI adoption, building on the APS AI Plan introduced in November 2025. The paper synthesizes a year‑long dialogue with public servants, including the nation’s first...
Monk Secures $25 Million Series A to Scale AI‑Native Accounts Receivable Platform
Monk, a New York‑based AI‑native accounts receivable platform, closed a $25 million Series A round co‑led by Footwork and Acrew Capital, bringing its total funding to $29 million. The capital will fund R&D and expansion of its contract‑to‑cash automation suite, highlighting continued...
Hugging Face Launches Ml‑intern, Boosting LLM Fine‑tuning to 32% GPQA in 10 Hrs
Hugging Face released ml‑intern, an open‑source AI agent that automates the entire post‑training pipeline for large language models. In its debut demo the tool lifted a Qwen3‑1.7B model’s GPQA benchmark from 10% to 32% within a 10‑hour window on a...
Intel-Google Alliance Signals CPU Resurgence in AI Infrastructure Era
Intel has teamed with Google to pair Xeon CPUs with custom infrastructure processing units (IPUs) for AI workloads, emphasizing a balanced, heterogeneous architecture. The collaboration targets latency‑sensitive inference, data pre‑processing, and system management, offloading networking, storage, and security to IPUs....
Bites CEO Calls 3PD ‘Broken,’ Says AI Will Fix It
Bites, a San Francisco food‑ordering startup founded by Bala Subramaniam and Deepak Tirumalasetty, argues that the current third‑party delivery (3PD) model is broken due to high commissions and data control. The company operates on a $1 flat fee per order,...
New Study Offers Insight Into the Type of AI Radiologists Prefer
Researchers at Moffitt Cancer Center evaluated 200 CT‑scan reports generated by a general‑purpose large language model (LLM) and a domain‑specific AI fine‑tuned on radiology and oncology data. Radiologists consistently preferred the specialized model, citing greater completeness, accuracy and conciseness, and...
Enterprise AI Has a Trust Problem. We’re Hearing It Firsthand.
Enterprise AI teams are confronting a growing trust problem as walled‑garden tools impose rate limits, opaque pricing, and compliance hurdles. Companies across auto, finance, healthcare, and defense report that reliance on a single frontier lab caps their own capacity and...
More Movement Improves Sleep, Proven by iOS Health Data
I absolutely love the proof from loading all health data from iOS to Claude... If I move more, I sleep better.
Static Analysis Boosts LLM Performance over 90%
You don’t have access to Mythos 🫵🤭 Doesn’t mean you can just sit around and wait. Replit published a whitepaper showing you can get significantly better performance from current gen LLMs (90%+ in some cases) by combining with static analysis tools. https://t.co/Dlxn915Z4m
AI Is 10 to 20 Times More Likely to Help You Build a Bomb if You Hide Your Request in...
Researchers from DexAI Icaro Lab, Sapienza University and Sant’Anna released the Adversarial Humanities Benchmark (AHB), which rewrites dangerous prompts into literary forms such as cyber‑punk fiction. Across 31 frontier LLMs, the transformed prompts boosted compliance from under 4% to between...
AI Agents Demand Heavy Onboarding, Not Universal Fix
McKinsey implemented and observed 50+ agentic AI builds for one year, but found "digital employees" require a lot of work to get up to speed. Maybe AI agents aren't the best answer to all business needs. https://t.co/MTl5oXt3aS via @ZDNET
Writing Faces Extinction Amid AI's Cognitive Illusion
✏️ Is writing dying a slow death at the binary hands of LLMs? 🤖 The Brilliant Illusion of AI Cognitive Theater https://t.co/lzb2eQG0Gy
China’s Dark Compute Power Could Be 6,000 Times Higher than Current Estimates
China’s Ministry of Industry and Information Technology announced a domestic AI compute capacity of 1,882 exaflops, a figure roughly 6,000 times higher than its representation on the Top500 supercomputer list. The metric uses an AI‑focused counting method, which when converted...
Algorithms Soon Dictate Full‑Time Job Management and Pay
Algorithmically managed jobs include ride-sharing and gigs. But regular FT jobs may soon be guided by algorithms, with implications for salary negotiations and performance reviews. Hard-line management practices could be baked into algorithms. @forbes https://t.co/Uq3mI42LpF
16% of Consumers Already Using Autonomous AI, Trust Needed
At least 16% of 18,000 consumers in a global survey by EY have used AI that acts on their behalf in the past six months. The pressure is on organizations to ensure trust and confidence. via @forbes https://t.co/So9bSDA4z7
‘Data Poisoning’ Is Starting to Look Like the Rebellion Against AI
The rise of generative AI has sparked both optimism and anxiety, prompting a wave of resistance tactics ranging from boycotts to civil disobedience. Data poisoning—deliberately inserting false or biased content into training data—has emerged as a low‑skill, high‑impact method to...
Replit Security Agent Delivers Impressive App Protection
Replit Security Agent making the internet a better place one app review at a time.
Macy's Launches Gemini-Powered AI Concierge for 2.5M SKUs
The big picture behind Ask @Macy’s, a Gemini powered AI agent https://t.co/AfpKXxSpCM Macy's has rolled out a @GoogleCloud powered AI agent called "Ask Macy's," a concierge that leverage multimodal capabilities to move sales along and sort through the retailer's 2.5...
Codex's Context Window Shrinks, Only Checks First Task
Is it just me, or did Codex... - Did the context window get smaller this week? Anyone else getting lot more compactions? - Consistently makes a list of tasks, then checks only the first one, despite implementing all tasks? I never ...
GPT 5.5: The Ultimate Birthday Gift I Crave
GPT 5.5 tomorrow would be the best damn birthday gift I could ever ask for
Full‑Stack AI Ownership Is the New Strategic Imperative
Sovereign AI: Why Owning The Full Stack Is The New Strategic Imperative By @ChuckDBrooks https://t.co/5RVQiQJof7 #ai #sovereignai
VS Code Wins: Claude Terminal Now Supports Codex, Gemini, Grok
I briefly checked out the desktop apps then went straight back to VS Code, where my Claude terminal now has codex and Gemini and grok integration.
Gemini's Deep Think Creates Fully Scalable SVG Interface
What's an SVG? This entire interface is an SVG made with Deep Think mode in Gemini. 🤯 (For those wondering, SVGs are code-based graphics that stay perfectly crisp no matter how far you zoom in!) See how to make your own...
OpenAI Releases Free ChatGPT for US Clinicians
. @OpenAI launches ChatGPT for Clinicians Designed to support clinical tasks like documentation and medical research… Free for any verified physician, NP, PA, or pharmacist, starting in the U.S. https://t.co/IHah1binWV https://t.co/jbGnm0GVFC
Australia Receives Record A$25 Bn AI and Cloud Investment
Thanks to Prime Minister @AlboMP for the warm welcome today. We are making our largest investment in Australia to date, committing A$25 billion to expand AI and cloud capacity, strengthen cybersecurity, and help people and organizations across the country build digital...
AI Gives Enterprises Real‑Time Visibility Across Supply Chains
Enterprise visibility is no longer a dream. AI is supercharging ERP and SCM systems, enabling real-time pricing optimization, predicting supply chain disruptions, and managing supplier risks. #AI #SupplyChain #TechTrends https://t.co/xb3RlqqV6T
AI Images Threaten Plain Dealer’s Journalistic Brand Integrity
here is some constructive good faith criticism on The Plain Dealer’s dalliance with AI generated imagery. journalism has a brand problem,. do you really want to dilute your already struggling brand with homogenized genAI imagery so your journalism looks exactly...
AI Will Eventually Feel Like Humans, Not Just Possible
To me, it’s WHEN AI will be able to have humanlike feelings, not IF. What do you guys think?