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:
Box Unlocks Agentic Data Extraction to Help Companies Dig up Insights Faster
Box announced the general availability of Box Extract, an AI‑driven data‑extraction service that taps third‑party large language models such as OpenAI’s GPT‑5.2, Google Gemini 3 and Anthropic Claude Opus 4.5. The platform offers two built‑in agents—Standard for rapid, cost‑effective capture and Enhanced for deeper reasoning—plus the ability to create custom agents. Integrated into Box’s Enterprise Advanced plan, the tool parses documents into structural components and surfaces key information for downstream workflows. Box positions the service as a way to turn unstructured content into actionable data across finance, government, insurance and legal functions.
Parloa Raises $350M to Make Enterprise Customer Experience Fully Conversational
Parloa, a Berlin‑based AI customer‑experience automation startup, announced a $350 million late‑stage funding round at a $3 billion valuation, led by General Catalyst. The round brings total capital raised to over $560 million within four years and includes existing backers such as EQT...
DianaHR Deploys Onboarding Agent to Cut HR’s Paperwork Gravity
HR-as-a-service provider DianaHR launched an agentic AI onboarding system designed to automate the paperwork‑heavy compliance layer of people operations. The platform works from Slack or email, ingesting offer letters, verifying signatures, and interacting with dozens of payroll and benefits tools...
‘Not Regulated’: Launch of ChatGPT Health in Australia Causes Concern Among Experts
OpenAI has rolled out ChatGPT Health to a limited Australian audience, positioning it as a tool that can interpret medical records and offer personalized health advice. The service is not classified as a medical device, meaning it bypasses mandatory safety...
‘It’s AI Blackface’: Social Media Account Hailed as the Aboriginal Steve Irwin Is an AI Character Created in New Zealand
An AI‑generated Instagram and Facebook persona called “the Bush Legend” portrays an Indigenous‑styled wildlife presenter, amassing roughly 90,000 Instagram and 96,000 Facebook followers. The character, created by a South African living in New Zealand and flagged by Meta, mimics Steve Irwin...
Higgsfield AI Doubles Revenue in Two Months
🚨 Lean AI Leaderboard Update A new speed record was just set. Higgsfield AI went from $100M → $200M run-rate in 2 months. Let that sink in. Here’s what makes this unprecedented 👇 • $200M run-rate in < 9 months...
AINA Introduces AI-Driven Hiring Platform Backed by $1M Raise
Cyprus‑based AINA secured a $1 million seed round from a private angel investor to scale its AI‑driven hiring platform. The solution automates repetitive recruiting tasks, offers AI‑based career coaching, and promises faster, more precise candidate shortlisting. To date, the platform has...
Anthropic Cuts AI Productivity Forecasts in Half After Analyzing Claude's Real-World Failure Rates
Anthropic’s fourth Economic Index Report reveals Claude’s success rate falls sharply as task complexity rises, with roughly 60% success on sub‑hour API requests and 45% on tasks exceeding five hours. The new data forced the company to halve its AI‑driven...
Innovations in Energy and Finance Are Further Inflating the AI Bubble
Innovations in energy storage and novel financing are accelerating AI development, but they also risk inflating a speculative bubble. New renewable micro‑grid projects aim to lower the massive electricity bills of data centers, while credit‑linked funding vehicles mimic the mortgage‑securitisation...
AI Security Platform WitnessAI Raises $58m to Expand Globally
WitnessAI secured $58 million in strategic funding led by Sound Ventures, aiming to accelerate global expansion and broaden its AI security suite. The platform provides enterprises with real‑time visibility and control over large‑language models and autonomous AI agents, addressing governance gaps...
Oversonic Robotics Announces Strategic Partnership with NantWorks and Opens New Los Angeles Office
Oversonic Robotics announced a strategic partnership with NantWorks and the opening of a 5,000‑square‑foot Los Angeles office. The alliance will accelerate research in both medical and aerospace domains, bringing the company’s first cognitive humanoid robots into U.S. clinical trials and...
ChatGPT Saved Me From a Roadside Nightmare — This Is Exactly What We Should Be Using AI for, Not for...
In a recent TechRadar opinion piece, a driver used ChatGPT’s multimodal capabilities to diagnose a mysterious 19 mph speed‑limiter warning on his dashboard. By uploading a photo, the AI instantly identified the feature, guided him to the correct button, and restored...
Slingshot Wins $27 Million Space Force Contract for AI Training System
Slingshot Aerospace secured a $27 million, 18‑month contract from the U.S. Space Force to advance its AI‑based training system, TALOS, under the Operational Test and Training Infrastructure (OTTI) program. TALOS acts as an autonomous virtual opponent, generating adaptive satellite maneuvers and...
How People Use ChatGPT – Analysis of OpenAI’s First User Study
OpenAI released its inaugural user study, "How People Use ChatGPT," detailing how individuals and organizations integrate the model into daily workflows. The research, blending surveys, anonymized usage data, and interviews, finds that roughly a third of interactions are work‑related, with...
Meeting the New ETSI Standard for AI Security
ETSI has released EN 304 223, the first globally applicable European standard that defines baseline security requirements for AI systems across enterprises. The standard aligns with the EU AI Act and introduces concrete provisions covering deep neural networks, generative AI,...
AI’s Energy Bottleneck Fuels Deflation, Pressures the Fed
AI is changing productivity, inflation, and also monetary policy 🤖📉 On @bitsandbips, @CampbellJAustin, @ramahluwalia & @perkinscr97 are joined by @DivesTech to unpack: ⚡ AI’s energy bottleneck 🏦 Pressure on the Fed ₿ Why Bitcoin enters the picture Timestamps: 🚀 00:00 Introduction ⚖️ 6:36 Why the DOJ probing...
Apple Notes Could Become My Go-To Life Organizer with This Big iOS 26.4 Upgrade
Apple is set to upgrade its Notes app in the upcoming iOS 26.4 release, embedding Google Gemini’s generative AI through Siri. The integration will let users generate notes, add shopping‑list items, and pull data from Safari or Mail without manual copy‑pasting....
PayZen Appoints Ken Brause as CFO to Support Next Phase of Growth
PayZen, an AI‑driven patient affordability platform, announced Ken Brause as its new chief financial officer. Brause brings finance leadership experience from fintech firms DailyPay and OnDeck, focusing on strategic finance, capital markets, and scaling fast‑growing businesses. His role will oversee...
Lawmakers Eye Dynamic Ecommerce Pricing
U.S. lawmakers are intensifying scrutiny of AI‑driven dynamic pricing, highlighted by Senator Chuck Schumer’s criticism of Instacart’s 2025 experiment that showed up to 23% price variation on grocery items. In Tennessee, House Bill 1468 would outlaw personalized algorithmic pricing by...
What Are The Remaining Bottlenecks For Humanoid Robotics? A Physical AI Study Finds The Limits Are Physical, Not Cognitive
A new Physical AI study reveals that humanoid robotics is limited more by physical constraints than by artificial intelligence. The research highlights data scarcity, the sim‑to‑real gap, energy and latency limits, and safety verification as the primary bottlenecks. It argues...
Aembit Announces Agenda and Speaker Lineup for NHIcon 2026 on Agentic AI Security
The NHIcon 2026 virtual conference, organized by Aembit, will explore the emerging security challenges of agentic AI in enterprises, featuring keynotes from Phil Venables, Misam Abbas, and Anthropic’s deputy CISO Jason Clinton. Sessions will cover practical topics such as LLM...
Tines Launches AI Interaction Layer to Unify Agents, Copilots and Workflows
Tines Security Services launched an AI Interaction Layer that unifies AI agents, copilots, and workflow automation within a single, secure interface. The platform introduces Model Context Protocol support, human‑in‑the‑loop controls, and built‑in auditing to address the fragmentation of AI tools....
Stop Wasting AI Investment on a Broken Change Approval Process
The article argues that AI investments in developer productivity are wasted when change approval processes remain heavyweight and batch sizes stay large. It promotes working in small batches and lightweight, automated approvals as the antidote to slow pipelines and compliance...
How Human Motion Is Fueling the Robot Revolution — by Teaching Robots Like Atlas to Move in Lifelike Ways
At CES 2026 Boston Dynamics showcased Atlas performing fluid, human‑like motions, from martial‑arts stances to precise squats. The company captures human movement using Xsens motion‑capture suits or VR controllers, then retargets the data to the robot’s unique kinematics. In simulation,...
Teamwork.com: AI-Powered Client Management for Modern Agencies and Consultants
Teamwork.com introduces an AI‑enhanced project management platform tailored for marketing agencies, consultants, and client‑facing teams. It consolidates tasks, approvals, time tracking, and resource planning into a single system, reducing fragmented communication and bottlenecks. AI‑driven workload and capacity tools help managers...
Advancing Human-Machine Interfaces with Memristive Technology
Researchers have unveiled a new memristive architecture that dramatically improves human‑machine interfaces (HMIs). The technology leverages analog resistance states to store and process data directly at the sensor level, cutting latency and power draw. Early prototypes demonstrate real‑time pattern recognition...
New RoboReward Dataset and Models Automate Robotic Training and Evaluation
Stanford and UC Berkeley researchers released RoboReward, a new dataset and benchmark for training vision‑language reward models in robotics. The dataset pairs real‑world robot videos with textual descriptions and graded success scores, enabling models to evaluate task performance automatically. Using...
MongoDB Combines Database and Embedding Models for Simplified AI Development
MongoDB announced the general availability of its Voyage 4 family of embedding models and the multimodal‑3.5 model, integrating them directly into Atlas and the community edition. The new models—voyage‑4, voyage‑4‑large, voyage‑4‑lite, and voyage‑4‑nano—offer varying trade‑offs in accuracy, latency and cost, while...
Teaching AI How to Forget
In this episode Ben Lorica interviews Ben Luria, CEO and co‑founder of Hirundo, about the rising importance of machine unlearning for enterprise AI systems. They explore how organizations can remove or forget specific data points from trained models to comply...
This Old Pennsylvania Coal Town Could Get a Reboot From AI
The former Homer City coal plant, once the largest in Pennsylvania, shut down three years ago, leaving the town without its 2‑gigawatt power source. A new natural‑gas facility is slated to begin operations in 2027, delivering 4.5 GW—more than double the...
Model Security Is the Wrong Frame – The Real Risk Is Workflow Security
Security teams are still focusing on protecting AI models, but recent incidents show the real risk lies in the workflows surrounding them. Malicious Chrome extensions harvested chat data from over 900,000 users, and prompt‑injection attacks can coerce AI coding assistants...
Sadiq Khan to Urge Ministers to Act over ‘Colossal’ Impact of AI on London Jobs
London Mayor Sadiq Khan warned that artificial intelligence could trigger "mass unemployment" in the capital unless the government intervenes. He cited research indicating that 70% of job skills will change by 2030, especially in finance, creative and professional services. City...
The Real AI Talent War Is for Plumbers and Electricians
AI firms are now battling for a scarce pool of electricians, plumbers, and HVAC technicians needed to build the power‑hungry data centers that drive artificial‑intelligence workloads. The U.S. faces an average annual shortfall of about 81,000 electricians, with McKinsey estimating...
AiMOGA Deploys Robotic Traffic Police Unit at Intersection in China
AiMOGA Robotics has placed its Intelligent Police Unit R001, a humanoid traffic‑policing robot, at a busy intersection in Wuhu, China, transitioning from pilot trials to routine urban use. The robot performs standardized hand signals and behavioral reminders for non‑motorized traffic...
AI and the Future of Warfare with US Under Secretary of War Emil Michael
In this episode, Under Secretary of War Emil Michael explains how the Department of War is deploying a generative AI platform, GenAI.mil, to over a million users within its first month, leveraging models like Gemini and Grok to boost decision‑making...
AI Agents Need a GTM Specialist, Not Just Sales Teams
"You can't just hand an AI Agent to the sales team or marketing team to deploy. Not if they don't have the skills to implement the agent, train it, manage it." with @kylecnorton CRO @kylecnorton "You have to find that GTM...
China’s Zoomlion Integrating AI and Embodied-Intelligence Robotics for Intelligent Manufacturing
Zoomlion Heavy Industry Science & Technology is rolling out a full‑chain AI system that links construction machinery, smart factories and humanoid robots. The company now operates 12 smart factories and more than 300 production lines, enabling high‑mix, small‑batch output with...
'Society Cannot Function if No One Is Accountable for AI' — Jaron Lanier, the Godfather of Virtual Reality, Discusses How...
Jaron Lanier, the virtual‑reality pioneer, argues that AI systems must remain under human accountability in the second episode of the new podcast *The Ten Reckonings*. He warns that delegating responsibility to machines would undermine law and civilization. The discussion references recent...
New Report From Meltwater and We. Communications Reveals 90% of PR Teams Are Integrating Generative AI Into Their Workflows
Meltwater and We. Communications released their inaugural State of PR Report, surveying over 1,100 communications professionals worldwide. The study finds that more than 90% of PR teams have incorporated generative AI into daily tasks such as drafting releases, content optimization,...
Auto-Close Tickets only when AI Reply Succeeds
Any @Zendesk experts out there? We have a gen AI reply out there for @haveibeenpwned which attempts to answer a support question, but has a fallback for if it can’t. How can we automatically close the ticket if a reply...
Q4 Wins Global AI Award for Finance Innovation
Q4 @q4tweets wins AI in Finance & Banking category in Global AI Awards Q4, the investor relations (IR) platform offering AI-powered tools for earnings communications, investor targeting, sentiment analysis. Corporate finance functions getting agentic capabilities https://t.co/6a7IvPvaGJ
Spain-Based DAO Xmaquina Raises $10M for Early-Stage Robotics Investments
Spain-based DAO Xmaquina has raised more than $10 million through its on‑chain Genesis Auction model, attracting nearly 2,000 contributors ranging from crypto‑focused venture funds to strategic angels. The auctions used uniform pricing and open participation, removing preferential terms typical of traditional...
Design Better AI Agents, Not Just Faster No‑Code
“No-code AI agents” ≠ easy AI. It really means: • intent over tools • workflows over models • memory over prompts • distribution over intelligence Agents aren’t built faster — they’re designed better. https://t.co/GTyMfcQMIB
Wikipedia’s Existential Threats Feel Greater Than Ever
Wikipedia marks its 25th anniversary amid mounting political, legal, and technological pressures. Conservative figures accuse the site of left‑wing bias while governments in the UK and Saudi Arabia propose age‑gating or imprison editors. AI bots scrape its content, driving traffic...
UiPath’s Platform Transition - a Case of Execution over Re-Invention. COO Ashim Gupta Explains
UiPath posted its first quarter of net‑new ARR growth in two years and achieved GAAP profitability, crediting execution discipline over new product launches. The company tightened renewal planning, deepened account management, and focused on post‑sale deployment to accelerate value realization....
Bain & Company Launches AI-Focused Venture Ecosystem With Leading Global VC Firms
Bain & Company announced the creation of an AI‑focused venture ecosystem in partnership with several leading global venture‑capital firms. The initiative will combine Bain’s strategic consulting expertise with VC capital to back early‑stage artificial‑intelligence startups. By establishing a dedicated pipeline,...
The CFO Case for Probabilistic Forecasting With AI | Bruno Annicq
In this episode, CFO Bruno Annicq of Wellhub shares how he transformed the company’s forecasting by adopting an AI‑driven probabilistic ensemble model, moving the finance team beyond single‑scenario planning. He outlines his EMPOWER planning framework, emphasizing cash discipline, sustainable growth,...
America Under Surveillance with Michael Soyfer
In this episode, Kevin Ball talks with Institute for Justice attorney Michael Soyfer about the rapid expansion of surveillance technologies such as automated license‑plate readers, facial‑recognition cameras, and predictive policing tools across U.S. municipalities. Soyfer explains the Fourth Amendment challenges...
Worldline Empowers Agentic Commerce with New AI Capabilities
Worldline announced two new capabilities—Model Context Protocol (MCP) servers and the ConnectAI developer hub—to make its Global Collect payment platform agent‑ready. The MCP acts as a secure bridge that lets large language models initiate payments, refunds, and status checks via...
DataMesh Launches DataMesh Robotics to Enable Industrial Embodied AI Training with Executable Digital Twins
DataMesh has introduced DataMesh Robotics, an embodied AI data product built on its executable industrial digital twin platform. The solution lets robot OEMs train, validate and evaluate robots in dynamic, process‑driven virtual factories, producing synthetic data and measurable reward signals....