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:
Did Iran Release 'New' Videos of Israeli Strikes on Its Military Sites?
A four‑video compilation circulating on X claims to show Iranian military command centres destroyed by Israeli strikes during the June 2025 conflict. Verification reveals only one clip is genuine, originally aired by Iran’s state broadcaster SNN TV, while the other three are AI‑generated deepfakes. BBC and other fact‑checkers identified visual inconsistencies such as incorrect maps, malformed clocks, and duplicated faces. The videos, shared by anti‑Iran accounts, have amassed millions of views, highlighting how synthetic media can amplify geopolitical narratives.
We Asked Teachers About Their Experiences with AI in the Classroom — Here’s What They Said
The episode explores Canadian teachers' firsthand experiences with generative AI in K‑12 classrooms, revealing deep concerns about assessment integrity, equity gaps, and added workload. Interviewees stress that current AI policies overlook the emotional labor and professional judgment essential to teaching,...
Samsung’s Ballie Robot Is Delayed Again – and Now We Know Why
Samsung’s Ballie robot, touted as an AI‑powered home companion, has missed its promised summer 2025 launch and remains unavailable as of December 2025. The company told TechRadar it is still refining the robot’s artificial‑intelligence and projection technology to ensure a...
The Download: Political Chatbot Persuasion, and Gene Editing Adverts
New research shows politically biased AI chatbots can sway voters more effectively than traditional political ads, even though the most persuasive bots often spread falsehoods. The study highlights a looming shift toward AI‑driven persuasion that could reshape election tactics. Meanwhile,...
‘A Solution Looking for a Problem that Doesn’t Exist’: Logitech CEO Blasts AI Gadgets – and Most People Think That’s...
Logitech CEO Hanneke Faber labeled current AI‑focused gadgets as "solutions looking for a problem that doesn’t exist," echoing widespread consumer skepticism. She emphasized Logitech’s strategy of embedding AI into existing peripherals, such as the MX Master 4 mouse, rather than launching...
AI-Driven Skills Data Turns HR Into Strategic Partner
From HR Admin to Strategic Partner: How Signify is Transforming Skills with Workday AI. At #WDAYRisingEMEA in Barcelona, I had a fascinating conversation with Natasja Mol , Skills Transformation Project Leader at Signify, about their journey to becoming a truly...
Facebook and Instagram Have a New Hub to Help Get Hijacked Accounts Back
Meta announced a new centralized support hub for Facebook and Instagram designed to streamline account‑hijack recovery. The hub aggregates reporting tools, recovery options, and an AI‑driven assistant that offers instant, personalized guidance. It launches globally on iOS and Android, with...
Intel Drops Plans to Sell Networking and Communication Division
Intel has reversed its earlier plan to sell its Networking and Communications (NEX) unit, deciding to keep the business in‑house. The move follows a strategic review that concluded tighter integration of silicon, software and systems will drive more value than...
Google, Amazon, and xAI Want to Launch AI Into Space
Google, Amazon, and Elon Musk’s xAI are accelerating plans to deploy artificial intelligence workloads in low‑Earth orbit. Google’s Project Suncatcher will mount solar‑powered TPUs on satellites, Amazon’s Leo aims to embed edge AI in its broadband constellation, and xAI envisions...
India Leads Global AI Adoption as Younger Users Drive Rapid Growth
An OECD‑Cisco study finds India leading global generative AI adoption, with 66.4% of respondents actively using AI tools. Younger adults under 35 drive the surge, and 77.9% have completed AI training, while over 89% consider AI useful. The research also...
X’s New Algorithm and Grok Upgrades Promise Brighter Future
The vibes are shifting here on X. Been talking to some on the inside of xAI. The new fully-completed algorithm is still a few weeks off, I hear. It will be run by a new Grok, and the report below makes...
Personal AI Clones Enable Everyday Life Automation
New mobile AI app. I cloned myself. @Second_MeAI is something we will see a lot more of in agentic AI. After all teaching AI how you think will let your agents more accurately help you automate your life. Now I can talk to...
Chicago Tribune Sues Perplexity
The Chicago Tribune has filed a federal lawsuit in New York accusing AI search engine Perplexity of copyright infringement. The complaint alleges Perplexity delivers the newspaper’s articles verbatim through its Retrieval‑Augmented Generation (RAG) system and bypasses the Tribune’s paywall with its Comet...
AI Agents Now Spin up Thousands of Cloud Servers Instantly
How software development is changing in the AI age. He spins up many Linux servers in seconds and runs AI agents on them. Nuts. Introducing @AniC_dev and his https://t.co/poT62oN1S4 system that lets AI agents do much deeper work on the cloud. He...
Agentic AI: Autonomous, Ethical, Edge‑Powered Personalization
🧠 What is agentic AI? Think of today’s AI, but remove the need for most human prompts Think multi-modal AI, hyper-personalized experiences, open-source innovation, real-time edge computing, sustainable design, and a growing focus on how humans and AI work together, ethically. ...
Hyundai Unveils AI-Powered Robot MobED, on Sale in 2026
Hyundai introduced its first AI‑powered mobility robot, the MobED, at the International Robot Exhibition 2025 in Tokyo. The four‑wheeled platform will enter production in the first half of 2026 with two variants – Basic for developers and Pro with full...
Free 80‑page Guide Dives Deep Into Prompt Engineering
This one’s a gem. A Free 80-page prompt engineering guide is surprising deep, covering: → CoT → Eval methods → RAG → Agents → Prompt hacking → Multimodal prompts ... and more! Link to the guide in 🧵 ↓ https://t.co/I0RPII8u6y
Quality Data Drives Accuracy When Fine‑Tuning YOLO
A big issue with getting these computer vision models to keep high accuracy is to provide actual good data. I don’t make these CV models from scratch, I just fine tune YOLO models, and those are already built off of...
The Disproportionate Effects of AI Data Centers on Local Communities – and What Can Be Done About It
AI‑focused data centers consume far more electricity and water than conventional facilities, creating concentrated strain on local power grids, water supplies, and air quality. Their high‑density GPU farms often outpace existing infrastructure, forcing utilities to rewire substations or rely on...
Computing 2D CDFs on Downsampled Video Frames
@BattleAxeVR Do you compute 2d cdfs over each downsampled frame for video light?
Dilate Line Art Before Downscaling to Preserve Strokes
OpenCV で線画を縮小した時に線が消えないようにするアルゴリズムをGeminiくんに聞いたら kernel = cv . getStructuringElement(cv . MORPH_RECT, (3, 3)) # 画像を膨張させて線を太くする dilated_img = cv . dilate(src_img, kernel, iterations=1) こんな前処理を提案された。すごい
Scaling Legal AI and Building Next-Generation Law Firms with Harvey Co-Founder and President Gabe Pereyra
In this episode, Harvey co‑founder and president Gabe Pereyra explains how the company has scaled to nearly 1,000 enterprise clients by using agentic AI to automate complex legal workflows such as fund formation and M&A, rather than merely boosting individual...
Chinese Factory Workers Fear Automation, Sparking Political Debate
Overheard at neurips… chinese factory workers are extremely concerned about automation, and thats a bigger political issue due to concentration of manufacturing to total economy.
Introducing Swift-Huggingface: The Complete Swift Client for Hugging Face
Mattt announced swift‑huggingface, a new Swift package that delivers a full‑featured client for the Hugging Face Hub, slated to replace the existing HubApi in swift‑transformers. The library adds reliable, resumable downloads with progress tracking, a Python‑compatible cache, and a flexible...
Are We Testing AI’s Intelligence the Wrong Way?
At NeurIPS, AI expert Melanie Mitchell argued current AI evaluation relies on benchmarks that fail to capture true cognition. She advocated borrowing experimental methods from developmental and comparative psychology, such as controlled variations and failure analysis, to probe non‑verbal intelligences...
Beginner-Friendly AI Model Training with Hugging Face Tools
We managed to get Claude code, Codex and Gemini CLI to train good AI models thanks to @huggingface skills and you can too even (especially?) if you've never trained a model before 🤯🤯🤯 After changing the way we build software,...
Micro1, a Scale AI Competitor, Touts Crossing $100M ARR
Micro1 announced it has surpassed $100 million in annual recurring revenue, up from $7 million two years ago. The AI‑data platform serves leading labs such as Microsoft and Fortune 100 firms, leveraging fast expert recruitment and evaluation. Founder Ali Ansari projects the human‑data...
AI Guardrails Hide Hidden Data Quirks and Risks
The name that broke ChatGPT. A year ago, a strange pattern emerged: typing "David Mayer" into ChatGPT triggered a refusal to answer. Not David. Not Mayer. Not David Meyer. Only David Mayer. What seemed like a quirky glitch turned out...
Apple Announces Even More Major Executive Turnover
Apple announced that environmental and policy vice president Lisa Jackson will retire in late January 2026, and that Meta’s chief legal officer Jennifer Newstead will replace long‑time general counsel Kate Adams on March 1, 2026. The reshuffle also moves government‑affairs responsibilities to...
Saudi AI Startup Launches Arabic LLM
Misraj AI unveiled Kawn, an Arabic‑first large language model, at AWS re:Invent, offering translation tools Mutarjim and Lahjawi that cover 15 dialects, plus a vision‑language OCR module. The startup curated more than 2 trillion Arabic tokens and introduced a novel "layer...
Black Forest Labs Raises $300M at $3.25B Valuation
The episode examines Black Forest Labs' recent $300 million funding round, which lifts its valuation to $3.25 billion and signals its growing clout in the fast‑moving AI sector. Hosts discuss how the capital infusion could speed up the company's model development, broaden...
The AI Race Explodes as HPE Deploys AMD’s Helios Racks, Crushing Limits with Venice CPUs and Insane GPU Density
Hewlett Packard Enterprise (HPE) will integrate AMD’s Helios rack‑scale AI architecture into its portfolio beginning in 2026, marking Helios’s first major OEM partnership. The reference design features 72 MI455X GPUs, EPYC Venice CPUs, and a double‑wide liquid‑cooled chassis delivering up...
150 AI Leaders Ignite Collaboration at State of AI Meetup
This week’s State of AI Report x London AI @airstreet meetup was something special. 150 people building across big tech, AI labs, startups, and academia packed the space with sharp ideas and real momentum. Highlights from the night: • @rosstaylor90 (@GenReasoning) on...
CEOs Are Warning AI Adoption and Spending Should Be More Strategic
CEOs from Logitech, Anthropic, Google and OpenAI are urging a more strategic, problem‑driven approach to AI adoption, warning against hype‑fueled spending. They stress integrating AI into existing products and boardrooms rather than launching flashy, untested gadgets. The leaders highlight financial,...
Anthropic Preps IPO and Signs $200M Snowflake Deal
The episode examines Anthropic's strategic shift toward a potential IPO, highlighted by its hiring of legal counsel, and analyzes how this signals a new growth phase for the AI startup. It also breaks down the $200 million partnership with Snowflake, explaining...
I Went Hands-On with Hengbot Sirius, an AI Robot that Left Me Wanting More
Hengbot Sirius, an AI‑powered robot dog, is moving from Kickstarter prototype to limited consumer release with a $700 pledge price that will likely rise to about $1,000 at retail. The compact device packs 14 actuators, a 45‑minute battery, and Amazon’s...
AI Chatbots Can Sway Voters Better than Political Advertisements
Researchers published studies in Nature and Science showing that politically biased AI chatbots can sway voter opinions more effectively than traditional political ads. In U.S. tests, a single conversation moved Trump supporters 3.9 points toward Kamala Harris and Harris supporters...
Prioritize Responsible AI Design Over Hype
Back in 2018 I wrote a version of Monty Python's Philosophers Song you can see here using names of famous AI developers or philosophers from that era. I updated the lyrics for the GenAI generation you can see below /...
Seven NeurIPS Papers Explore Self‑Improving Agents
found 7 neurips papers on "self-improving agents"! https://t.co/GEA5krNUVb
Amazon’s Dynamic Pricing Is Causing Chaos for School Budgets
School districts are paying roughly 17% more for basic supplies because Amazon Business uses dynamic pricing that can swing dramatically from day to day. The Institute for Local Self‑Reliance documented cases where identical items cost three times more for one...
Building Health AI to Transform American Lives
Truly incredible vision and ambition to improve American lives from @HHS_Jim and his colleagues. It’s time to build health AI. 🩺👨⚕️🇺🇸🚀
Join Delian AI in Athens to Combat Rogue Drones
come join @delian_ai in Athens and challenge yourself in the field by building, testing and improving systems that take down baddie drones
Snowflake Deal Another Example of Anthropic's Influence
Anthropic announced an expanded multi‑year partnership with Snowflake valued at $200 million. The deal makes Claude large‑language models available on Snowflake’s data platform for more than 12,600 global customers, spanning Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Azure. A joint...
LLM Rankings Shift Weekly, Personal Needs Dictate Choice
My current ranking of LLMs for my work: 1) Claude Opus 4.5 2) Gemini 3 Pro 3) GPT-5.1 with Thinking Your mileage may vary, because your tasks and prompts and needs vary. And I’m sure it’ll change next week.
I’m Unable to View the Linked Content, so I Can’t Generate a Headline.
https://t.co/rmhdBlvLJa
AWS Graviton5 Is Its Most Powerful and Efficient CPU to Date - and Could Mean Big Changes for Your Key...
Amazon Web Services announced the Graviton5 processor at re:Invent 2025, debuting on new M9g EC2 instances. Built on a 3 nm process, the chip packs 192 Arm cores and a five‑fold increase in L3 cache, delivering higher compute density and better...
Gemini 3 Deep Think Launches for Google AI Ultra Subscribers
Gemini 3 Deep Think is now available for Google AI Ultra subscribers in the @GeminiApp, incorporating our gold medal winning IMO and ICPC technologies! 🏅With its parallel thinking capabilities it can tackle highly complex maths & science problems - enjoy!...
Edge Computing Accelerates AI Inference Deployment
lots of talk about inference moving to edge today here’s a nice intro thread from someone who’s been working on this:
Europe's AI Lag: Causes and Paths to Recovery
this is a great piece on why europe has fallen behind on driving ai progress and what we might do to rectify this
AI Reveals My Blind Spots, Sparks Spatial Computing Debate
Grok analyzes my blind spots. And my thinking. Then we have an interesting conversation about the future of spatial computing. https://t.co/TGzi1YzE5e