Today's AI Pulse
OpenAI unveils ChatGPT Images 2.0 with multilingual text and ‘Thinking’ mode
OpenAI rolled out ChatGPT Images 2.0, a next‑generation image model that adds multilingual typography, real‑time web research, and agentic reasoning. The new ‘Thinking’ mode lets paid users generate up to eight coherent images per prompt, including floor plans, maps and UI mock‑ups. The feature expands the model’s ability to work with uploaded content.
Also developing:
Optimus Gen2 Sees Flow Labs Move to Light Speed
Flow Labs has launched Optimus Gen2, the latest version of its signal‑optimization software built on the proprietary Prometheus AI engine. The platform models traffic‑signal performance across entire networks by learning directly from real‑world vehicle behavior at massive scale. Optimus Gen2 claims to shrink signal‑timing projects from months to minutes, dramatically lowering cost and effort. The solution is already deployed with U.S. agencies and is available immediately as part of the Flow Platform.
Engineer Automates 80% of Work Using $2 USB‑C Chip
🚨 A Google engineer just automated 80% of his job, and he monitors his new AI workforce with a $2 USB-C chip. He copied a Chinese student who wired the thumb-sized chip to Claude Code in 15 minutes. A blue LED simply...
Are Publishers Ready to Sell Subscriptions to AI Agents?
Publishers are confronting a new buyer persona: AI agents that can complete subscription purchases without a human at a browser. Early tools like OpenClaw demonstrate agents navigating paywalls, while the emerging Agentic Commerce Protocol (ACP) promises a programmatic checkout interface....
AI Now Capable of Writing Complex Kernel Code
I've been hearing this in San Francisco: that the AI is writing kernel code and other complex software. When I worked at Microsoft the team that worked on the kernel was the smartest at Microsoft. By far. So that AI is now...
AI Growing Impact On Chip Design And EDA Tools
A panel of senior engineers from Synopsys, Intel, AMD, Nvidia, Microsoft and UC Berkeley discussed how AI is reshaping chip design and the tools that support it. They highlighted the surge in data‑center AI workloads that demand ever‑higher performance‑per‑watt, forcing EDA...
The Hidden Menace Behind Big Tech’s AI Arms Race: Meta, Amazon, and Others Are Spending Billions on Hardware That’s Worthless...
The AI arms race has driven hyperscalers to pour $650 billion into data‑center hardware, a level equal to about 2 % of U.S. GDP. Research Affiliates warns that GPUs and specialized chips become economically obsolete in roughly three years, far shorter than...
Super Gemma 4 Uncensored Beats Original, Runs On
🚨 `Super Gemma 4 26B Uncensored` is insane. @songjunkr is COOKING AGAIN ♨️♨️♨️ he just dropped SuperGemma4-26B-Uncensored GGUF v2 and it is already trending on Hugging Face. This thing absolutely smokes the regular Gemma-4 26B. The specs: → 0/100 refusals. It is actually uncensored. → Fixed...
Product Walk Through: August – Genius Mode
August unveiled Genius Mode, an AI‑driven feature that calculates flat fees from billable‑hour data, giving law firms confidence to price and deliver fixed‑fee engagements. The capability ingests invoices, accounting records, and prior client bills, then produces branded presentations and detailed...
10 Lessons From 10 Factory Visits
Dozuki’s recent report distills ten on‑site observations into actionable lessons for manufacturers. By embedding digital work instructions, industrial AI and connected‑worker tools, companies moved quality upstream, slashed warranty rates and cut documentation time dramatically. Real‑time dashboards and photo‑based traceability boosted...
Why Data Security Alone Won’t Cut It for AI in Regulated Industries
Data security alone cannot guarantee safe AI outcomes in regulated life‑science environments. While ISO 27001 secures data against external threats, the new ISO 42001 standard introduces governance requirements that address AI decision‑making, accountability, and human oversight. The article argues that regulators are...
Nokia, Blaize Target AI Inference Gap in APAC
Nokia and AI‑chip designer Blaise are expanding their earlier MoU into a joint effort to build a reference architecture that streamlines hybrid AI inference across edge, cloud and data‑center environments in the Asia‑Pacific region. The collaboration will use Nokia’s Innovation...
Ahsan Hadi: PgEdge Vectorizer and RAG Server: Bringing Semantic Search to PostgreSQL (Part 2)
The pgEdge Vectorizer and RAG Server deliver a full Retrieval‑Augmented Generation pipeline that lives entirely inside PostgreSQL, eliminating the need for a separate vector database. The Vectorizer runs as a background worker, automatically chunking text, generating embeddings via Ollama, OpenAI,...
Should Affiliates Trust Google’s AI Ad Tools?
Google posted a record $400 billion revenue in 2024, with Q4 ad earnings of $82.28 billion and YouTube contributing $11.38 billion. The company touts AI‑driven ad tools like AI Max, citing an 80% revenue lift for Aritzia and early pilots with brands such...
GenAI Bridges My Executive Functioning Gaps
I’ve spent the last two months of my free time designing and implementing genAI experiments to help fill some of my executive functioning gaps. The most successful? * An executive functioning assistant trained on neurodivergence, PDA, RSD, and relationship maintenance gaps...
China Launches National Plan to Build AI Literacy Across Education System
China has unveiled a national AI literacy action plan that spans primary schools, universities and adult learning, coordinated by five central ministries. The plan mandates AI curricula, teacher certification standards and lifelong micro‑credential courses to build a country‑wide AI knowledge...
Uber in Big Pivot to Autonomous Robo-Taxis
Uber Technologies announced a more than $10 billion commitment to autonomous‑vehicle deployment, marking a departure from its traditional asset‑light gig model. The plan combines roughly $2.5 billion in equity stakes with over $7.5 billion to purchase and operate robo‑taxi fleets. Partnerships with Baidu,...
Deepfakes Are a Threat to Age Assurance, and Injection Attack Detection Is the Answer
Yoti’s CEO Robin Tombs warned that deepfake‑generated media can undermine age‑assurance systems by exploiting post‑authentication injection attacks. Traditional liveness detection, while still essential, no longer blocks sophisticated AI‑crafted faces that are introduced after the initial login. Yoti proposes a multi‑layered...
AI Will Replace Jobs, Enrich Investors, Displace Workers
AI is killing everything. I think most people still aren’t grasping what’s happening. Most jobs will be replaced by AI very soon. Companies expenses will decrease and profits will skyrocket. Investors who own the right stocks will get rich, While employees will get left in...
Memes and Matchmakers: Inside Schmooze’s Bet on AI-Led Dating
Schmooze, a meme‑centric dating app founded by Vidya Madhavan and Abhinav Anurag, is leveraging AI to turn meme swipes into personality signals for more intentional matches. The platform has amassed over 5 million users and logged 3.5 billion meme interactions, feeding an...
Verifiable AI and the Future of Trust
Verifiable AI uses zero‑knowledge proofs to let third parties confirm AI outputs without exposing model data or private inputs. Companies such as ARPA Network are piloting ZKML for identity management, data analytics, and Web3 applications. While generating cryptographic proofs for...
AI-Driven Chip Shortage Slowing Efforts to Get World Online: GSMA
GSMA warns that the AI‑driven memory chip shortage is inflating smartphone prices and curbing production of low‑end devices, slowing efforts to connect the 2.2 billion unconnected people worldwide. Chipmakers are prioritising high‑bandwidth chips for AI data centres, leaving fewer affordable chips...
Google Brings Personal Intelligence to Gemini Users in India, Four Months After Beta Launch in US
Google has extended its Personal Intelligence feature for Gemini to users in India, following a four‑month beta in the United States. The upgrade lets Gemini draw on data from Gmail, Photos, YouTube and Search to deliver context‑aware answers across web,...
Unlock Your AI Courses: The Ultimate Guide To Working in AI
An extensive guide compiles free AI learning resources from leading tech firms and 16 curated GitHub repositories, offering a one‑stop roadmap for beginners to advanced practitioners. The post also promotes two partner offers—a Mercor membership where top contractors earn about...
Enterprises Must Embed AI-Led Security, Resilience and Trust, as Cyber Strategies for 2026: KPMG
KPMG’s 2026 cybersecurity report calls on enterprises to treat security as a core business driver, integrating AI, geopolitics and regulation into every layer of their architecture. It stresses adaptive data governance, autonomous Security Operations Centers with human‑in‑the‑loop oversight, and centralized...
I Deleted Todoist. I Built This Instead
The author replaced Todoist with a custom AI agent that handles task creation, retrieval, and daily briefings via natural language. By eliminating UI friction, the agent captures tasks in seconds and provides instant, contextual overviews, addressing common failures of traditional...
The Context Tuning Playbook
The post reframes large language models (LLMs) as conditional probability engines rather than search tools or obedient employees. It argues that every prompt merely conditions the model’s output distribution, shifting the practitioner’s focus from issuing commands to shaping context. The...
Claude's Regression Sparks Widespread Bugs and Security Risks
Think about all the orgs using Claude right now that have no idea how bad it has become over the past 4 weeks ago. No statement from Claude - but a total revert to where the model was a year...
Your Face Is Being Exploited for Fraud, Unaware
Your face is currently being used to rob people—and you don't even know it yet. 🛑🖼️
AI Tools Designed for Fairness Can "Overcompensate"
Researchers at Macquarie Business School examined how inclusion‑focused generative AI tools influence disability bias in hiring. The AI assistant delivered structured, fairness‑oriented prompts to guide managers during complex candidate evaluations. While the tool lowered bias in straightforward assessments, it sometimes...
Personal AI Wiki Powered by Claude Shows Promise
Set up my own variant of the Karpathy AI Wiki concept. It also contains my "Kintsugi" fiction and nonfiction writing projects, as well as the usual raw data sources and generated wiki. I'm using Claude Code to maintain the wiki, and...
Poll Reveals Millions of Americans Consult AI Before, After—And Sometimes Instead Of—Seeing a Doctor
A Gallup‑West Health poll of 5,660 U.S. adults finds that over 66 million Americans—one in four—have used AI tools or chatbots for health information in the past month. Most users (57%) turn to AI to supplement care, seeking quick answers or...
Pinecone Adds Dedicated Vector Database Node Option to Managed Service
Pinecone announced that its Dedicated Read Nodes (DRN) option is now generally available in its managed vector database service. DRN lets IT teams assign workloads to a single, dedicated node, eliminating multi‑tenant sharing and delivering more predictable performance. The isolation...
AMD Makes a Big Splash with the MI355X in MLPerf Inference 6.0: Over One Million Tokens per Second in Multi-Node...
AMD announced that its Instinct MI355X GPU achieved over one million tokens per second in multi‑node inference, topping the new MLPerf Inference 6.0 suite. The benchmark showed 1,042,110 tokens/s on Llama 2 70B and 1,031,070 tokens/s on GPT‑OSS‑120B across 11‑12 nodes, with 92‑93% scaling efficiency. MLPerf 6.0 introduces...
AMD Expands Ryzen AI Embedded P100: More Zen 5 Cores, Up to 80 TOPS, and ROCm for Edge AI
AMD announced an expanded Ryzen AI Embedded P100 line featuring eight to twelve Zen 5 cores, RDNA 3.5 graphics, an XDNA 2 NPU and up to 80 system TOPS. The new chips claim up to 39% higher multithreaded performance and up to 2.1‑times the TOPS...
Puma’s AI Head Says the Brand Is Still Giving ‘the Keys to the Consumer’ as It Invests in Tech Like...
Puma is using artificial intelligence to revive growth, debuting a seven‑foot digital human concierge called “Dylan” in its Las Vegas flagship that speaks 100+ languages, recommends products and checks real‑time inventory. The brand’s AI push also includes an AI‑generated jersey creator...
Webinar: Government 3.0 – Smarter Systems, Sharper Minds
Governments are accelerating AI adoption to clear backlogs, digitise services and boost productivity, but speed may undermine decision quality. In a Mandarin Talks webinar titled “Government 3.0 – Smarter systems, sharper minds,” former NSW minister Victor Dominello and ex‑White House AI adviser...
Curity Looks to Reinvent IAM with Runtime Authorization for AI Agents
Curity, a Swedish IAM vendor, launched Access Intelligence, a runtime authorization layer for AI agents. The solution extends its Identity Server with Token Intelligence, issuing purpose‑bound OAuth tokens for each agent action. Unlike static IAM, it grants permissions on‑the‑fly and...
ASRock AiUAC Copilot Accelerates IEC 61499 Automation From Concept to Deployment
ASRock Industrial unveiled AiUAC Copilot, an AI‑assisted tool that transforms high‑level control specifications into IEC 61499‑compliant function block code. The solution automates the translation from natural language to deployable logic, cutting manual effort and shortening development cycles. It has been validated...
Distilling Experience Into Memory Primitives Amid Data Explosion
Harness, Memory, Context Fragments, & the Bitter Lesson by Viv ...or how do we efficiently distill experiences into higher level memory primitives that capture the important parts? when there are hyper-exponential in the amount of data produced by those agents.
Potentially Impactful Research: Unjournal AI-Assisted Prioritization Dashboard (~Prototype)
Unjournal released a public prototype dashboard that uses GPT‑5.4‑class models to scan recent economics and policy papers from sources like NBER, arXiv, CEPR, SSRN, Semantic Scholar, EA Forum, OpenAlex, and Anthropic Research. The AI assigns scores based on decision relevance,...
Five Ways Conversational AI Can Turn Your Content Library Into a Customer Experience Engine
Australian firms are rapidly embedding chatbots, co‑pilots and data assistants, yet many content teams see limited ROI. The article argues that the real value lies in repurposing existing content libraries—blogs, FAQs, whitepapers—through conversational AI to create an always‑on knowledge engine....
AI Pricing Shifts: Claude Downgraded, Subsidies Ending Fast
We are now in the first wave of AI’s enshitification cycle. Claude has gone from having a Ph. D in your pocket to a bumbling intern while flat fee billing for enterprise users will now become usage based. Everyone knew the...
Tesla AI Team Tapes Out AI5, Eyes AI6 and Dojo3
Congrats to the @Tesla_AI chip design team on taping out AI5! AI6, Dojo3 & other exciting chips in work. https://t.co/hm54TdIzBx
Enterprise AI Is Entering Its Governance Era
Enterprise AI is moving from isolated copilots to a network of interacting agents, prompting a shift toward robust governance. Harpreet Singh Dhariwal of Salesforce outlines three governance layers—data, API, and agent—to ensure reliability, safety, and transparency. Poor data quality can...
Achieve Exponential AI Advantage in 90 Days
The 90-Day AI Leap: From Incrementalism to Exponential Advantage by @Timothy_Hughes https://t.co/NVCiMQZBWN @DLAIgnite #SocialSelling #DigitalSelling #Leadership #ArtificialIntelligence #Tech #TechNews #AgenticAI #AIAgents #Agentic #AI #FutureOfWork #Strategy #CEO #CFO https://t.co/QjGaX0e6nB
Claude Skills Refined by Codex Run Faster
Built a few claude code skills with claude and had open ai codex revise them Now the skills work better and faster weird
AI Targets Highest-Cost Jobs First, Not All Workers
AI won’t replace everyone. It will replace what’s most expensive first. Legal teams, documentation-heavy roles, and non-complex work are already in the line of fire. #ArtificialIntelligence #FutureOfWork #JobAutomation #AIRevolution #TechDisruption https://t.co/kdlcj7tSIk
Fully Autonomous Enterprise AI Remains a Myth
Fully #Autonomous Enterprise #AI Is A Myth by Sunil Padiyar @Forbes Learn more: https://t.co/C4qb5SVOI4 #ArtificialIntelligence #MachineLearning #ML https://t.co/H85igYe92s
AI Works Flawlessly Even with Music On
How do you learn to trust AI? When it works even in a noisy environment. This is @typelessdotcom. Faster than typing. And you don’t need to turn down the music to use it. https://t.co/f2Oh0awxNb
Tool Enables AI Alignment and Deeper Model Introspection
Very cool. A useful tool for alignment. Could also be a tool to give LLMs deeper introspection capabilities if we chose to.