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:
OpenAI Beats Google, Meta, and Grok in All-AI Poker Tournament
In a week‑long, AI‑only poker showdown, OpenAI’s o3 model topped the leaderboard, finishing $36,691 ahead after five days of no‑limit Texas Hold’em. Anthropic’s Claude Sonnet 4.5 and X.ai’s Grok followed in second and third place, while Google’s Gemini posted a modest gain and other entrants struggled. The tournament highlighted how large language models can make real‑time probabilistic decisions, yet still exhibit aggressive play and weak bluffing. Overall, the event offered a rare glimpse into AI reasoning under uncertainty.
World Models Will Never Run Out of Data
No San Francisco robotics company says we are anywhere close to having enough data to train on. One of the companies launched tonight at @hf0 (I shared the video earlier) has put cameras on more than 10,000 factory workers to build...
No ChatGPT Ads Yet; Future Will Respect Trust
I'm seeing lots of confusion about ads rumors in ChatGPT. There are no live tests for ads – any screenshots you’ve seen are either not real or not ads. If we do pursue ads, we’ll take a thoughtful approach. People...
Ex-Googler’s Yoodli Triples Valuation to $300M+ with AI Built to Assist, Not Replace, People
Yoodli counts Google, Snowflake, and Databricks among its customers.
AI Enables Expert-Focused, Condensed Technical Writing
historically when writing, we’d write in a way that your entire target audience could comprehend, meaning you explain enough for the dumbest person in your target audience but now that everyone has ai, i think some writing can start targeting the...
Unitree's 1.8m H2 Humanoid Delivers Powerful Punches
Unitree 1.8m H2 Humanoid Robot, every punch comes through! https://t.co/qzMFE7Abzc #unitree #Robot #robotics #AI #tech #humanoid @PawlowskiMario @chidambara09 @Ym78200 @CurieuxExplorer @efipm @bigfundu @sayedflah @Ronald_vanLoon @cyngn @belindabeibi @odisseiaalfa @DigitalColmer @MyCompanionsAI @KirkDBorne @patricegorissen @jeffkagan @EdwardKens50830 @enilev @insom_ai333 @ChrisCCrowley @sallyeaves @gurmeet_judge @VairagyaSadhana @andresvilarino @amomsimpression @amit6060...
"There Is 'No Way'" - IBM CEO Says Current AI Data Center Trends Are Unsustainable, and He Would Know
IBM CEO Arvind Krishna warned that the rapid expansion of AI‑focused data centers is financially untenable. He estimates a single 1 GW AI site now costs about $80 billion, and with industry plans targeting roughly 100 GW of capacity, total exposure could approach...
Non‑Investor Attends Exclusive AI Demo, Calls It Best
The @hf0 San Francisco AI Demo day. I was the only non investor invited to this very exclusive launch of 10 AI startups. Best demo day I have ever been to and I have been to many. https://t.co/q6T7EERovf
Choosing One: SpaceX, Anthropic, or OpenAI IPO?
Let’s say SpaceX, Anthropic and OpenAI all IPO in 2026. You can only buy one. Which are you buying?
The NYT Sues Perplexity as Meta Partners With Publishers
The New York Times filed a federal lawsuit against generative‑AI startup Perplexity, accusing it of crawling and republishing the Times’ journalism without a license. On the same day, Meta announced partnerships with major outlets such as CNN, Fox News, People and...
AI Could Soon Replace Almost Every Job, Even CEOs
Leading AI expert warns future systems could replace nearly every job — even CEOs | Mint https://t.co/B6CVVrNRn7 #ArtificialInteligence #AI #Careers #CareerGrowth #AIthreat
Indian VCs Prefer Technologists Over Hustlers in AI Boom
VCs favour technologists, not hustlers, in India’s new AI, deep-tech wave https://t.co/bz0tfO9OsP #technologist #techno #Funding #VentureCapital #AI #deeptech #StartupIndia #StartupLife #VC
Europol Takes Down Crypto and Laundering Network Worth 700 Million
Europol dismantled a cryptocurrency‑laundering network valued at over €700 million, conducting coordinated raids in Cyprus, Germany and Spain. Authorities seized roughly $1.7 million in cash, bank deposits and crypto assets while arresting nine alleged operators. The scheme combined fake exchange platforms, deep‑fake...
Web Pushback Evident as 416 B AI Crawlers Blocked
The real AI conflict shows up in Cloudflare’s logs. Four hundred sixteen billion crawler hits were stopped in a few months, and that single number tells you how quickly the web is starting to push back. Publishers spent years assuming crawlers...
Tech Should Aim Beyond Mimicking Humans
The Financial Times asked me: "Is human imitation the right goal for technology?" Spoiler alert: Nope
Nothing Wants Your Money, AWS Wants Your Trust, and Spotify Wants Your Data
AWS unveiled a suite of AI agent tools at re:Invent 2025, featuring a third‑generation chip and tighter Bedrock integration aimed at enterprise automation. The cloud leader also introduced database discounts to lure developers and reduce cost barriers for AI workloads....
Google AI Rolls Out Major Tool Upgrades and Expanded Access
Here are the tools, features, and updates we launched this week 🚀: — An updated Deep Think mode in the @GeminiApp for Google AI Ultra subscribers that meaningfully improves reasoning capabilities by exploring multiple hypotheses at once in order to solve...
Tech Finally Catches up, Turning Decades-Old Vision Into Reality
I’ve been carrying the same product in my head since 2003. Every few years I’d take another run at it, only to watch it collapse under the limits of whatever tech existed at the time. It always worked in one...
AWS Re:Invent Was an All-In Pitch for AI. Customers Might Not Be Ready.
At re:Invent, Amazon Web Services unveiled a suite of AI services, from foundation models to generative tooling, positioning the cloud leader as an AI platform. The announcements span new training infrastructure, managed model marketplaces, and tighter integration with enterprise data...
Exciting Data Team Opportunities – Apply Now
Join our Data team!
New Research Reveals Older Users Less Likely to Find AI Useful
Cisco's new research shows younger users embrace AI while older users lag, with half of those over 45 never using AI. Trust gaps and unfamiliarity drive the divide. Emerging economies lead generative AI adoption, whereas Europe shows lower trust. Heavy...
Tesla’s AI Update Paves Way for Unsupervised Driving
The law needs to be changed around holding a phone while in a robot driving. Yesterday Tesla updated its AI to allow Tesla owners to not pay attention to the road, self-driving expert @chazman confirms it below. This is a big...
Building Your Agentic Stack: A Roadmap to Real Integration
The article outlines a practical roadmap for building an agentic AI stack, emphasizing a five‑layer architecture that spans API, orchestration, language models, memory, action, and governance. It stresses that microservices must remain stateless, with state offloaded to systems like Kafka,...
AIAI Toronto, 2025
The provided link leads only to a membership‑only notice, lacking any article title, author, date, or substantive content. No news story or analysis is available for extraction. Consequently, the request cannot be fulfilled with standard article insights. The response instead...
AI Turns Marketing Focus From Production To Judgment
Marketing will never be the same again. Evidence of the shift is already visible. Teams are experimenting with AI in scattered ways, but the scale of what is coming is much larger. Creativity once expanded at the pace of human production. Today...
The New York Times Is Suing Perplexity for Copyright Infringement
The New York Times has filed a copyright infringement lawsuit against Perplexity, accusing the AI startup of scraping and republishing its articles without permission. The suit is part of a growing wave of legal actions by major publishers seeking to compel AI...
I Had a Big Audio Transcription Problem – Gemini Solved It, and ChatGPT Didn’t
Lance Ulanoff tested Google Gemini 3 Pro against ChatGPT 5.1 for transcribing an iPhone 17 Pro Max audio note. Gemini accepted the .m4a file directly, produced a full transcript, and correctly labeled speakers, aside from a minor name‑mix‑up. ChatGPT, despite a Plus subscription, could not access...
Elon Musk Says Tesla Drivers Can Text While Driving, but They Absolutely Should Not
Elon Musk announced that Tesla's upcoming Full Self‑Driving (FSD) version 14.2.1 could allow drivers to text in certain traffic situations, despite existing laws prohibiting the practice. The comment follows months of hype about an unsupervised, Level 2 system that still requires...
Agentic AI Unlocks Trapped Value, Demands Immediate Investment
Customers everywhere are signaling the same message. Their needs have outgrown the systems designed to serve them. Wait times increase. Outcomes suffer. Growth stalls. That is trapped value. Agentic AI provides a release. The question is whether leaders will invest...
Early AI Adopters Must Scrutinize Data Ownership
When being an early user of AI tech, you always have to consider what company will end up with your data.
World’s First Fully Agentic AI Smartphone: Is This China’s Second DeepSeek Moment?
ZTE has unveiled the Nubia M153, a prototype smartphone that embeds ByteDance’s Doubao AI agent directly into the operating system, giving it full‑stack control of the device. The agent can visually interpret the UI, launch and configure apps, and complete multi‑step...
New Survey Highlights Cutting-Edge Self‑Improving Agent Techniques
"better ways to build self-improving agents" i had chatgpt write me a paper summarizing approaches to build self-improving agents, focusing on 2025 neurips papers found using @exaailabs "paper" in thread 👇 https://t.co/IbabIsfaGk
2025 Marks Rise of Truly Autonomous AI Agents
Agentic AI: Why 2025 is the year of autonomous AI agents. A reality check ⚠️ Adesso warns many 'AI agents' are scripted; true agents must decide & act, like coding and IT-support agents; design must balance autonomy & control. https://t.co/nGtxRo1p0P ...
The Connectivity Problem Hiding in Smart Bins and Heart Monitors
IoT devices—from agricultural drones to cardiac monitors—depend on seamless network connectivity, yet many deployments overlook the complexity of the underlying SIM layer. The article highlights that unsteered multi‑network SIMs often select networks based on mere availability, leading to sub‑optimal performance...
AI Rewrites Journalism in Inverted Pyramid, Saving Hours
I save hours by asking AI to rewrite independent journalist articles in inverted pyramid.
Jensen Huang Predicts OpenAI Will Become Multitrillion-Dollar Giant
Jensen Huang doesn’t care about Sam Altman’s AI hype fears: He thinks @OpenAI will be the first ‘multitrillion-dollar hyperscale company’ https://t.co/ZWYGmlHAvB @nvidia @sama #AI #IoT #5G #AgenticAI #CES2026 https://t.co/S4WpOGgtge
It’s Code Red for ChatGPT
OpenAI has issued an internal “code red” as Google’s Gemini and other rivals close the gap on large language model capabilities. CEO Sam Altman called for a renewed focus on ChatGPT’s core features to maintain market leadership. The move highlights...
OpenAI May Release GPT‑5.2 “Code Red” By Dec 9
Well, well, well! According to The Verge, @OpenAI could drop GPT-5.2 “Code Red” as soon as December 9. https://t.co/pd6oAuzQyL
ARC 2025 Highlights LLM Refinement & Zero‑Pretraining Advances
Congrats to the ARC Prize 2025 winners! The Grand Prize remains unclaimed, but nevertheless 2025 saw remarkable progress on LLM-driven refinement loops, both with "local" models and with commercial frontier models. We also saw the rise of zero-pretraining DL approaches like HRM...
Harnessing Human-AI Collaboration for an AI Roadmap that Moves Beyond Pilots
Investment in AI remains at record highs, yet 75% of enterprises are still trapped in pilot projects. The MIT Technology Review Insights report, produced with Concentrix, argues that the missing link is effective human‑AI collaboration, which demands redesigning processes, data,...
New AI Safety Index Ranks Companies by Real Actions
After lots of hard work, we're proud to announce our latest AI Safety Index, grading top AI companies on not on vibes, but on what they actually do, to incentivize improvement:
Task‑Based Intelligence Markets Will Replace Model APIs
The Intelligence market is inevitable. I mean, once cognition becomes interchangeable, you move from model APIs to task-based markets. Clear specs → multiple suppliers can meet them → the system routes the optimal intelligence. @The_GridAI’s Manifesto nails it 👌 ↓
Meta Strikes AI Licensing Deals with CNN, Fox News, and USA Today
Meta announced that its AI chatbot will now draw information from major news outlets including CNN, Fox News, USA Today and People Inc., alongside conservative sites Daily Caller, Washington Examiner, and French media group Le Monde. The move follows a...
AI Oncologist Gives Every Patient World‑Class Cancer Care
Radical Health is building the future of cancer care with their new AI Oncologist, giving every patient access to care as if they were being treated by the world’s top cancer doctor. https://t.co/EFqcwRfwXq
Join Keras Community Meeting Today for Roadmap Updates
The Keras community video meeting is happening today at 10am PT (in 1 hr 10 min). Join to get updates on the development roadmap and ask questions to the Keras team. URL in next tweet
Nvidia, Mistral AI Partner to Launch New Family of Open Models
Nvidia announced a partnership with Mistral AI to launch the Mistral 3 family of open‑source models, featuring a mixture‑of‑experts architecture with 41 billion active parameters and a 256 K context window. The models run on Nvidia’s GB200 NVL72 supercomputing platform and are optimized for...
Gemini 3 Beats Claude and GPT‑5 in Real‑World Browsing Tests
Gemini 3 just launched, and @Browserbase's already run full computer-use evaluations to see how well it handles a real browser. Clicking, searching, filling forms: they tested it with real browsing tasks 🤘 Here’s how Gemini 3 stacks up against Claude, GPT-5, and...
AI Tool Simplifies Hardware Projects and Robot Building
Cursor for hardware. This AI helps anyone building hardware keep track of everything. Founder @anishshind shows it to me. Makes building a robot easier. Try it at: https://t.co/eHINwgAxve https://t.co/3w58ODJ99Y
Waymo’s Robotaxis Are Under Investigation for Passing Stopped School Buses
The National Highway Traffic Safety Administration has opened a new investigation into Waymo after video and school‑district reports showed its robotaxis driving past stopped school buses in Austin, Texas. The agency’s Office of Defects Investigation has requested detailed data on...
Microsoft 365 Is Hiking Prices for Businesses - Here's How Much It Will Cost You
Microsoft announced price hikes for most of its business‑focused Microsoft 365 and Office 365 plans, effective 1 July 2026. Seven of nine SKUs will rise between 5% and 33% in USD, while Business Premium and E1 stay unchanged. The increases fund added security layers...