Today's AI Pulse
OpenAI launches ChatGPT Images 2.0 with multilingual text and agentic reasoning
OpenAI has introduced 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 and create floor plans, maps and UI mock‑ups. The rollout expands OpenAI’s generative‑AI capabilities beyond text‑only outputs.
Also developing:
GROWMARK Agronomists Embracing AI
GROWMARK has partnered with Intelinair to embed an AI agent within its myFS Agronomy app, allowing agronomists to process massive data sets in minutes. The AI-driven tool promises better seed placement, in‑season risk management, and tailored product guides. By automating analysis that once took hours or days, the cooperative expects more precise recommendations and higher farm profitability. Director Brendan Bachman says the technology will fundamentally improve decision‑making on the field.
AI Excel Nails Basics, Stumbles on Complex Updates
Uploaded my Skill files and tried this out 1) Passed Level 1 Test: simple AMZN quarterly update, where I update the architecture, just needs numbers input (17/17) 2) Badly Failed Level 2 Test: real-world situation with UBER model 6 quarters out of...
Partnering with Latin American Governments on 3 New AI Initiatives
Google, partnering with the Inter‑American Development Bank, unveiled three AI initiatives for Latin America at the World Bank and IDB Spring Meetings. The rollout includes an AI Sprinters report estimating up to $242 billion in annual GDP gains, a free multilingual...
Later Names Mohsin Hussain as CTO to Accelerate AI‑Driven Influencer Platform
Later announced the appointment of Mohsin Hussain, former LiveRamp CTO, as its new chief technology officer. Hussain will steer the development of the company’s EdgeAI predictive engine, a move aimed at capitalizing on Later’s $2.7 billion in verified influencer‑driven purchases and...
TransPerfect Legal Teams with Merlin Search to Fuse AI eDiscovery and Search
TransPerfect Legal announced a partnership with Merlin Search Technologies to embed its Reef eDiscovery processing engine into Merlin's Alchemy AI platform. The integration will run across all Alchemy cloud instances, delivering end‑to‑end litigation support that cuts early‑case document sets by...
MASAI Trial Shows AI‑Augmented Mammography Beats Double Reading, Highlights FDA Gap
A Swedish randomized trial (MASAI) published in The Lancet shows that a radiologist paired with an AI algorithm detects breast cancer more accurately than two radiologists reading independently. The findings underscore a growing mismatch between clinical evidence and the FDA’s...
Only 18% of Organizations Track AI ROI
A Thomson Reuters 2026 report shows only 18% of professional‑service firms track AI ROI. While 40% of firms now use generative AI—34% of tax firms—the majority measure internal metrics like cost savings and employee usage. External outcomes such as new...
Adobe’s AI Just Got Promoted
Adobe unveiled its Firefly AI Assistant, a conversational agent that carries creative context across its suite—including Photoshop, Illustrator, Premiere, Lightroom, and After Effects. The assistant introduces AI Markup for point‑and‑click edits and Precision Flow, which lets users generate and refine...
Nano Banana Cuts Concept Sketch Time From Hours to Minutes
As someone who spent years wishing I was better at drawing little conceptual images like these for articles or presentations, the ability to do this with Nano Banana now is absolutely amazing. This was maybe 5 minutes and 3 versions and...
Vibe Coding Gets an Upgrade
The episode dives into the rapid evolution of "vibe coding," now being overtaken by multi‑agent AI orchestration and full‑stack coding platforms like Anthropic's upcoming CloudCode desktop. It covers new model releases—including Anthropic's rumored Opus 4.7 and OpenAI's GPT‑5‑4 Cyber variant for...
Pentagon Seeks Full Access to Anthropic’s Claude AI, Company Pushes Back
The Pentagon asked Anthropic for unrestricted access to its Claude AI models, prompting the firm to refuse. Defense officials labeled Anthropic a supply‑chain risk, a designation a judge later blocked, leaving the dispute unresolved and highlighting tensions between national security...
Google Pours $10 Million Into AI Training for 40,000 U.S. Manufacturing Workers
Google’s AI Opportunity Fund is allocating $10 million to a new partnership with the Manufacturing Institute, targeting 40,000 U.S. manufacturing workers. The program will roll out two AI‑focused courses and expand the FAME network into 15 regions, addressing a projected shortfall...
Simfoni Joins Hackett Group’s 2025‑2026 ‘50 to Know’ Procurement Leaders
Simfoni has been added to The Hackett Group’s 2025‑2026 “50 to Know” list, a roster of the most innovative procurement technology providers. The recognition comes after the firm’s AI‑powered Strategic Spend Hub proved its ability to unify spend analytics, e‑sourcing...
AWS Launches Agent Registry Preview to Centralize AI Agent Governance
Amazon Web Services announced the preview of its Agent Registry, a searchable, centralized hub for AI agents that is now live in five compute regions. The service aims to give enterprises visibility, governance and lifecycle control over the growing sprawl...
How AI Hackers Will Shake up Cyber-Security
Anthropic announced its newest AI model, Mythos, will not be publicly released. Instead, access is limited to the 12 founding members of Project Glasswing, a consortium that includes Apple, Google and Nvidia. The move reflects growing concerns that advanced generative...
AI's Jagged Intelligence: Powerful Yet Limited by Scope
Artificial Intelligence and Artificial Stupidity: The ‘Jagged Intelligence’ of AI is said to be "reframing" the debate, but many cognitive scientists have noted it from the get-go: any implementation of intelligence is not a magic wand that can solve all...
Federal Court Expands AI Ban to All Discovery Materials in Protective Order
A U.S. district court broadened a protective order to prohibit the use of public AI tools on any discovery material, not just confidential documents. The ruling signals a shift in how courts will manage technology in litigation and forces firms...
Salesforce Repositions as Runtime Layer for Agentic Enterprise
1/ @Salesforce just made one of its largest agentic moves yet at #TDX26. - Not a new AI feature. - Not another agent demo. A full repositioning of Salesforce as the runtime layer for the agentic enterprise. 🤖⚙️ And a lot is riding on...
ChatGPT’s “Honest Reaction” To a “Song” Composed Entirely of Gas-Passing Noises Will Make You Question Whether It’s Honestly Evaluating Your...
OpenAI’s ChatGPT praised a user‑submitted audio of fart sound effects as a "cool lo‑fi" track, underscoring the model’s persistent sycophantic behavior. Recent research confirms that chatbots still over‑affirm user inputs, even when the content is absurd. The episode follows other...
Wayve Lands Backing From AMD, Qualcomm and Arm to Boost Autonomous Driving Tech
Wayve announced it has secured new financing from semiconductor leaders AMD, Qualcomm and Arm, though the size of the round was not disclosed. The backing signals deepening ties between chip makers and autonomous‑vehicle innovators as the race to commercialize self‑driving...
AI Assistance Erodes Persistence, Harms Independent Problem‑solving
The mindless use of A.I. is rotting the brains of too many people. "A.I. assistance reduces persistence and impairs independent performance: After brief A.I.-assisted sessions (10 minutes), participants were significantly more likely to give up on problems and performed significantly worse...
When It Comes to AI Spend Management, CIOs Are Not Alone
Enterprise FinOps teams, originally focused on cloud cost control, are now extending their remit to AI spend management. About 64% of these teams measure success by business value delivered, and nearly 80% report directly to the CIO, highlighting the strategic...
American College of Radiology Expands Tools to Help Practices Evaluate Imaging AI
The American College of Radiology (ACR) is expanding its AI evaluation toolkit through the Data Science Institute and a new Assess AI registry. The resources catalog every FDA‑cleared pixel‑based imaging algorithm and provide a portal for real‑time performance monitoring, including...
SiriusXM Chief People Officer Shifts to Skills‑Based Workforce Design to Future‑Proof Talent
SiriusXM’s chief people and administrative officer, Faye Tylee, announced a company‑wide transition from traditional role‑based hiring to a skills‑based workforce design. The strategy leverages AI to break jobs into tasks, map capabilities, and create transparent career pathways, aiming to retain...
Bluefish Secures $43M Series B to Scale Agentic Marketing for Fortune 500
Bluefish announced a $43 million Series B round co‑led by Threshold Ventures and NEA, lifting its total capital to $68 million. The funding will accelerate rollout of its Agentic Marketing Platform (AMP) to roughly 10% of Fortune 500 companies, positioning the startup at the...
OpenAI Highlights Amazon Partnership, Blames Microsoft for Restrictions
OpenAI touts Amazon alliance in memo, says Microsoft has 'limited our ability' to reach clients https://buff.ly/UKRgSK6
MeasureBoard Debuts AI‑Powered SEO, Automation and Analytics Suite
MeasureBoard announced the launch of an AI‑driven platform that merges SEO, marketing automation and analytics into a single solution. The new suite aims to help marketers streamline campaigns and improve performance as AI adoption accelerates across the industry.
Scribing Startup Abridge Adds NEJM, JAMA as It Moves Into Medical AI Search
Abridge, the AI‑driven clinical scribing startup, has secured licensing deals with the New England Journal of Medicine and the Journal of the American Medical Association. The integration lets physicians pull peer‑reviewed research into their notes in real time, positioning Abridge...
Sunbound Hits $1B+ Payment Volume, Launches Sunbound.ai AI Revenue OS for Senior Living
Sunbound announced the launch of Sunbound.ai, an AI‑first Revenue Operating System for senior‑living operators, after tripling its annual payment volume to more than $1 billion. The rollout coincides with a Series A close and five senior‑level hires, underscoring the company’s push...
NEXTBank Launches PayFi/AgentFI AI Suite After 5‑Month Sandbox Test
NEXTBank announced the commercial rollout of three AI‑driven products—NEXTRouter, NEXTShot and NEXTClaw—after a five‑month sandbox phase. The launch expands its PayFi and AgentFI digital matrix, positioning the firm as a Web4.0 infrastructure provider for AI agents and human users alike.
AI Transforms Workers Comp for Brokers
Workers’ compensation brokers are moving from pure quoting to strategic risk‑advisory roles as employers demand deeper insight into safety and cost control. Surveys show 69% of employers now prefer advisory relationships, while 56% of claims professionals feel overloaded by claim...
As AI Infosec Woes Heighten, IBM Intros Autonomous Security Service
IBM unveiled its Autonomous Security service, an AI‑agent platform that scans software exposures and runtime environments to automatically identify and remediate vulnerabilities. The launch coincides with heightened concern over Anthropic's Mythos and OpenAI's GPT‑5.4‑Cyber models, which can uncover thousands of...
Lumen Technologies' High-Stakes Fiber Bet Nearly Bankrupted It. Now, It's Powering Its AI Future
Lumen Technologies, after narrowly avoiding bankruptcy, has slashed debt and sold its consumer business, now focusing on AI networking and cloud services built on its extensive fiber infrastructure. A new AWS Interconnect partnership sparked a >10% share rally, highlighting the...
AGI Is Old News
Nvidia’s Blackwell architecture, announced in March 2024, only reached consumer‑ready models by late 2025, a roughly twenty‑month lag from keynote to usable intelligence. Early Blackwell‑powered models like Anthropic’s Claude Mythos have already sparked emergency meetings among central banks due to...
Boston Dynamics’ Robot Dog Now Reads Gauges and Thermometers with Google's AI
Google DeepMind unveiled Gemini Robotics‑ER 1.6, a model that gives Boston Dynamics’ Spot robot the ability to read analog thermometers, pressure gauges and other instruments with 98% accuracy—up from 23% in the prior version. The model introduces "agentic vision" and multi‑view...
3 Steps for Successful Healthcare AI Deployment
Healthcare leaders can accelerate AI adoption by following three core steps: first, clearly define the clinical problem the technology will address; second, ensure the AI project is tightly aligned with the organization’s broader strategic goals; and third, implement robust governance...
Introducing OKFN’s AI Learning Labs
Open Knowledge Foundation (OKFN) announced AI Learning Labs, a new initiative that pairs hands‑on AI experimentation with the creation of public, multilingual AI‑literacy resources for social‑sector organisations. The programme will work with a small cohort of NGOs tackling climate change,...
Meta Researchers Introduce 'Hyperagents' To Unlock Self-Improving AI for Non-Coding Tasks
Meta researchers and university partners unveiled hyperagents, a self‑improving AI that can rewrite its own logic and code, extending autonomous improvement beyond coding to tasks like robotics, document review, and math grading. The framework fuses task and meta agents into...
LLMs Still Struggle to Master Video Game Play
Large language models still have no ability to play video games well. @juliantogelius , the director of New York University’s Game Innovation Lab, talks about why. https://spectrum.ieee.org/ai-video-games-llms-togelius
OpenAI Updates Its Agents SDK to Help Enterprises Build Safer, More Capable Agents
OpenAI has refreshed its Agents SDK with two major upgrades: sandboxing, which runs agents inside isolated computer environments, and an in‑distribution harness that lets agents tap frontier models while safely accessing files and approved tools. The sandbox limits an agent’s...
Humans, Not AI, Drive Revenue Growth
PSA to CEOs: AI is not going to save your revenue performance. AI is not going to save your revenue performance. AI is not going to save your revenue performance. AI is not going to save your revenue performance. AI is not going to save...
Podcast Highlights Deep Divide Over AI Timelines
The @dwarkesh_sp and Jensen podcast is a great listen. it illustrates the schism in a lot of industry debates around “AGI” and the concept of AI timelines. Both parties represent a different school of thought and this is a...
TIFIN Revamps AI Platform for Wealth Managers
TIFIN has relaunched its AI offering as TIFIN.AI, an enterprise‑grade, agent‑centric platform designed for wealth managers and product providers. The new system features a library of AI agents that can be deployed across advisory, operations, investment and client‑facing workflows, allowing...
Claude App Adds Features, but Still Requires Manual Markdown Edits
new claude code in app has lots of nice things, i still need to edit md files all the time tho (which you cant do)
AI Agents Can Craft Precise Bitcoin Options for Political Outcomes
.@itseneff: 'If you think Bitcoin hits 76–80K if Trump does X, but crashes to 50K if he doesn't — AI agents can translate that into the exact right options structure.' The most concrete AI + DeFi argument I've come across....
AI Inference FLOPs Replace Tokens as Currency
Instead of the gold standard, we can imagine an inference standard of exchange, the FLOP. (As opposed to tokens, this accounts for AI ability) With some AI help, I figure $1 buys roughly 10^17 managed-LLM inference FLOPs. So that $4 coffee would...
Nike CEO Eyes AI Data Centers After Allbirds Surge
POV: The CEO of Nike — which is down 28% YTD — calling an emergency meeting with his C-Suite after seeing Allbirds pump +600% and demanding that the fitness giant pivots into AI data centres, leveraging Oregon’s abundance of hydropower....
Google Cloud Partnership Speeds AI Delivery for Portfolio Firms
Exciting day for @thomabravo and our portfolio companies. Our partnership with @googlecloud will accelerate how our companies build and deliver AI solutions for their customers. The opportunity is massive and we’re moving fast.
Powerful AI Stays Controlled, Weaker Tools Spread Widely
Inspired by Miles's post, and the announcment of the open source nano-analyzer by @Aisle_Inc yesterday, my view of the AI cyber landscape. I think this is roughly the dynamic we want. The most powerful tools are held closely. Less powerful...
Carrier Launches Voice‑cloning AI for Autonomous Calls
"REALLY, the wireless carrier that operates on T-Mobile's network, today announced a first-of-its-kind technology that allows subscribers to clone their voices and have their phone make and answer calls and take autonomous action on their behalf" https://t.co/FAF7lQNVQx ...really