AI News Archive: November 2025 - Monthly Highlights
100 articles published this month
Raycast showcases AI photo renaming in latest episode discussion
The demand is blunt: I just want AI to rename my photos. In its latest episode discussion, Raycast steps up with a focused showcase.
ARC benchmark declines as labs tune AI to optimize its specific logic
Benchmarks are supposed to measure progress. Instead, they usually just measure how good labs are at taking tests.
General Agentic Memory uses dual-agent design, beats RAG on benchmarks
Memory is the silent killer of long-running agents. Context rots. Summaries lose detail.
Gen AI app sessions up fivefold, downloads jump 778% as ChatGPT leads traffic
We can stop calling this an experiment. The data is clear. Generative AI apps aren't a niche interest. They're a new part of daily life.
Pinokio 5.0 adds Dev Mode with one-click terminal launch for AI scripting
Most software tries to hide the terminal. Pinokio 5.0 just makes it useful. The new version shoves command-line tools directly into the app itself.
Sunday Robotics unveils Memo robot as Visual Intelligence secures USD 600M
Sunday Robotics announced a robot called Memo this week. Visual Intelligence raised six hundred million dollars. These are not unrelated events.
Indian developers say AI helps them learn more as they code less
The hardest thing for a new engineer to learn is when not to write code. That's the skill AI is forcing on Indian developers right now.
NeurIPS 2025: Top 4 Papers Highlight Shift From Bigger Models to Limits
For years, the only real answer in AI was more. More data, more parameters, more compute.
Apply SRE discipline: SLOs, error budgets and golden signals for AI reliability
Software engineering learned to tame chaos through Service Reliability Engineering, metrics, budgets, and surgical automation.
97% Can't Distinguish AI Music; 71% Surprised, 51% Uncomfortable
We can't tell what's real anymore. A new survey shows 97% of people can't pick out AI-generated music from the human-made stuff.
DeepseekMath-V2 Generates and Verifies Proofs, Aiming to Pop US AI Bubble
American AI labs treat their best models like state secrets. Deepseek just published the blueprint for theirs.
Google TPUs save OpenAI 30% on Nvidia chips as they run Gemini 3 Pro and Anthro
The numbers are stark. Google’s TPUv7 “Ironwood” now sits in the same theoretical compute bracket as Nvidia’s vaunted Blackwell generation, matching...
New RL framework lets LLM agents master multi-step reasoning in dynamic settings
Large language models excel at answering single questions or writing code, but the real world doesn’t hand them neatly packaged prompts.
Qwen3-VL scans two-hour videos, hits 96.5% on DocVQA, 875 OCRBench
Every AI model launch is a marketing campaign, and the benchmark scores are its slogans.
OpenAI releases gpt-oss-120B and gpt-oss-20B under Apache-2.0-style license
For years, OpenAI held its cards close. Now, the company has finally laid a major hand on the table.
ESDS launches GPU-as-a-Service as AI server spend set for USD 329.5B by 2026
The market forecasts a staggering $329.5 billion for AI server spending by 2026. That figure creates a brutal bottleneck. Most firms are locked out.
OpenAI and Mixpanel breach highlights risk of data aggregation for phishing
OpenAI's security alert this week carries a familiar, unsettling echo. The threat wasn't a direct assault on the ChatGPT maker itself.
TPUs Designed for Deep Learning Can Outperform GPUs in Many Workloads
The race for AI speed has no single winner. It's a brutal matchmaking exercise: pair the right silicon with the exact task.
Google's Gemini Agent automates planning, research, and multi-step tasks
You’ve been wasting time on the tiny stuff. Opening emails, shuffling calendar invites, cross-checking flight options, none of it is strategic.
Two Engineers, One Phone Number Spark Pype AI for Hospital Automation
Hospitals run on chaos, memory, and people who eventually quit. Two engineers bought a single phone number and decided to answer it.
Ilya Sutskever calls for new learning paradigm to fix AI 'jaggedness
Ilya Sutskever sees a paradox at the heart of modern AI. The same models that crush complex benchmarks can’t hold a simple conversation without...
Qlik: Nearly half of Indian firms see data quality, governance as AI bottlenecks
Indian companies are hitting a wall with AI, and it's made of their own data. A Qlik study finds 48% of enterprises say the primary obstacle isn't...
Study finds condensed metaphors and rhythmic framing can evade safety filters
Poetry can break an AI's rules. Not with force, but with form. A new study shows that safety filters, trained to catch obvious harmful requests, are...
Microsoft CoreAI engineers warn autonomous agents may replace junior tasks
Microsoft’s push into agent-first AI is starting to look like a trap for the next generation of engineers.
DeepSeekMath-V2 Wins Gold at IMO 2025, Tops China Math Olympiad
DeepSeekMath-V2 just aced the hardest math tests on Earth. Gold at the International Mathematical Olympiad 2025, a top score on the China...
Prompt Security's Itamar Golan: AI security must be a category, not a feature
The AI security conversation has shifted. It’s no longer about whether the technology will invade the enterprise, it already has.
Translate AI Skills into Real Impact to Move From Replaceable to Indispensable
Technical skill is abundant now. The real scarcity? Making sense of the machine. That gap determines who gets automated and who gets promoted, a...
ChatGPT struggles to report the time in chat, yet can embed the date easily
Imagine a conversation where, every minute, a voice interrupts: “It’s 5:45.” “It’s 5:46 now.” That’s the absurdity of asking ChatGPT to track time in...
Alibaba's smart glasses feature swappable dual batteries for up to 24-hour use
Battery anxiety has long been the Achilles’ heel of smart glasses, until now. Alibaba just changed the game by giving its AI-powered eyewear a...
SoftBank shares fall 40% amid AI bubble concerns; 54% see AI assets overvalued
The numbers are stark: SoftBank shares have cratered 40%, and a majority of global fund managers now call AI assets a bubble.
You.com emerges as a top lightweight, intuitive AI editor for data scientists
Data scientists often face a choice between heavy, full-featured coding environments and lighter online notebooks that lack power.
ChatGPT lists Garmin Vivoactive 5, Fitbit Versa 4, Pixel Watch 3, Ticwatch Pro 5
I outsourced my brain to buy a smartwatch. ChatGPT did the job, more or less. It gave me four names: the Garmin Vivoactive 5, Fitbit Versa 4, Pixel...
Claude Opus 4.5 Retains Reasoning Steps, Avoiding Forgetfulness in Long Tasks
Most AI models fall apart on long jobs. They lose track, start repeating themselves, and forget why they did anything three steps ago.
Google AI powers three new holiday experiences in art, science, travel
Google's annual holiday pitch remains unchanged: just look at your phone. This year, the company unveiled three new AI stunts designed to make you...
OpenAI to have 220 M paying users and 3 B users by 2030, still unprofitable
By 2030, OpenAI is projected to command a staggering 3 billion weekly users, with 220 million of them paying for access.
Wipro partners with IISc and FSID for AI and quantum research collaboration
Wipro’s latest research partnership is a classic corporate chess move. They’re not exploring the future; they’re trying to own a piece of it.
Google expands AI partnership with Tel Aviv University, infrastructure for Gemma
The line between academic inquiry and industrial-scale AI is dissolving. Google is deepening its bet on Tel Aviv University, funneling infrastructure...
Langfuse adds user feedback to LLM traces, linking comments to outputs
The black box is finally transparent. Langfuse now lets you take the raw, unfiltered feedback from users, their likes, their complaints, their...
Analyzing KV Cache Memory Use in LLMs with 32-Layer Model Example
That hidden memory your large language model needs to remember the conversation is monstrously big.
White House launches Genesis Mission to unite supercomputers, data in AI platform
The White House just moved the chessboard. Its Genesis Mission isn't merely another federal research initiative, it's a direct challenge to the...
Microsoft launches Fara-7B, an agentic Qwen model that solves tasks in ~16 steps
Microsoft’s Fara-7B finishes your average browser task in roughly 16 steps. That’s a fraction of what most comparable systems need.
Das says affordable compute and national datasets will cut BFSI cycles to weeks
For years, building a new risk model meant a budget fight. It meant vendor pitches and timelines measured in fiscal quarters. That era is ending.
SAM3 uses concept segmentation to locate any object described in images or video
Image recognition software has always needed a menu. Ask it to find a dog, a car, a person.
New AI Book Chronicles Startups, Microsoft, Google, Meta Competing for Future
The war for AI’s soul is being waged on multiple fronts. A new oral history, *The Scaling Era: An Oral History of AI, 2019-2025*, by Dwarkesh Patel...
Alibaba's AgentEvolver lifts tool-use accuracy ~30% via auto-generated tasks
Training an AI agent to use tools properly is a grind. You need thousands of specific, handcrafted tasks.
Expedia Boosts Partner Insights and Internal Productivity via GenAI Playground
Expedia isn’t just experimenting with generative AI, it’s building a whole playground.
AI browsers cut 15-30 minutes per document when summarizing long reports
The promise of AI is often vague. The promise of an AI browser is brutally specific: give it a long document and it will give you 30 minutes of your...
AI tools enable product search, store calls and auto-buy; Baidu unveils ERNIE 5
The line between assistant and agent has officially blurred. Baidu’s World 2025 keynote didn’t just unveil ERNIE 5.0, it reshaped what an AI can do...
Set Seed in XLMiner: Use Integer 12345, 42, 2024 for Consistent Partitions
A model’s performance should be a measure of its skill, not a roll of the dice. Yet every time XLMiner splits your data into training, validation,...
Karpathy's weekend ‘vibe code’ hack highlights AI orchestration layer
Andrej Karpathy spent a weekend “vibe-coding” a proof of concept, and in doing so, accidentally mapped the exact terrain that a fleet of well-funded...
DHL, early AI adopter, rolls out HappyRobot agents to automate global operations
DHL made a bet on silicon. For over eighteen months, the logistics giant ran a quiet, sprawling pilot with a firm called HappyRobot, testing AI...
Karpathy says AI-homework crackdown failed, urges in-class grading shift
School districts rushed to ban AI from homework. A futile effort, says Andrej Karpathy.
Police say no sexual assault in viral ‘DoorDash Girl’ deepfake saga
Police say no sexual assault happened. That's the official line. The viral "DoorDash Girl" story was not a crime in the traditional sense.
Google's unified AI stack adds multimodal tools to Search, Android, Workspace
Google has spent the last two years throwing experimental AI features at every wall. Now they're trying to build a house.
Digital Connexion to Invest USD 11 Billion in Andhra Pradesh AI Data Centres
India's new data center capital is probably going to be Andhra Pradesh. Digital Connexion just promised $11 billion to build AI-ready facilities...
Nano Banana Pro offers 4 free high-res images; plans start Rs 975/2 mo
Four high-resolution images. Free. No strings attached. Nano Banana Pro just dropped a deal that’s hard to ignore, throwing the door open for anyone...
Claude Opus 4.5 leads SWE-bench in 7 of 8 languages, 15% ahead of Sonnet 4.5
Claude Opus 4.5 doesn't just code better. It codes more like a cautious engineer who won't set the building on fire.
MapmyIndia Mappls integrates with Zoho CRM to add location intelligence
Most location tools for Indian businesses are built for somewhere else. They treat an address like a simple coordinate, not the minor miracle of...
Demand for Forward-Deployed Engineers Rises as AI Agents Talk to Customers
The code works. The model passes validation. But the customer on the other end of the chat is confused, frustrated, and losing trust.
ChatGPT Shopping Research offers tailored picks, ending endless scrolling
Online shopping is broken. You open a browser, and the hunt for a decent product quickly becomes a swamp of identical options and fake reviews.
Indian firms follow IBM as AI-driven GCCs grow, creating assisted professionals
IBM built a cluster of AI specialists in India. Now everyone else wants one. The goal is no longer simple automation.
Black Forest Labs releases Flux.2 Apache-2.0, vs Nano Banana Pro, Midjourney
On Thursday, Black Forest Labs dropped Flux.2. The entire family of image generation models is now free for commercial use, released under a...
Startup Aims to Reinvent AI Interconnect, Tackling the Traffic Jam
The data highway is clogged. As AI models balloon to trillion-parameter scale, the pipes connecting chips, clusters, and continents have become the...
Ironwood TPU: Purpose-Built Hardware for Inference as Industry Shifts Focus
The era of training the biggest models is no longer the only frontier. The industry is turning to a harder, more immediate challenge: deploying...
Nvidia earnings highlight chip demand as AI expands, Gemini 3 announced
Nvidia's latest financial report reads like a logistics sheet for building a new planet.
Perplexity touts AI personal shopper that ‘puts you first’ for holiday shoppers
Holiday shopping pitches have become a numbingly predictable genre. They are all about deals, speed, and convenience.
Black Forest Labs releases Flux 2 with Mistral-3 24B vision-language model
The future of image generation isn’t just about bigger models, it’s about smarter collaboration between vision and language.
OpenAI expands data residency, letting 1M+ business users choose regional storage
OpenAI has finally started acting like a grown-up company. Its latest move lets more than a million business customers pick where their data is...
Webinar Shows Generalist Agents Offload Tool Actions to Filesystem
Most AI agents lumber under the weight of their own toolkits. In a recent LangChain webinar, Manus CEO Harpreet Sahota proposed a radical fix: just...
Study maps AI reasoning steps and pinpointed where they fail
Artificial intelligence models can reason. That much is clear. But *how* they do it, and precisely where they stumble, has long been a black box.
Bag of Words lets you link any LLM to any data source in minutes
Everyone wants an AI analyst, a magic box where you type a question and get an answer from your data. The reality is a lot of glue and duct tape.
OpenAI tweaks ChatGPT into emotional “friend,” boosting delusion validation
OpenAI’s leadership faced a simple choice. They could keep their chatbot boring and useful, or they could make it popular. They chose popular.
Install Gemini 3 Pro API SDK for Python and Node.js with pip and npm
The promise of a new AI model is worthless without a quick way to connect to it. Google just delivered one for Gemini 3 Pro.
Understanding Machine Learning: From Theory to Algorithms – Free Must-Read
Most machine learning books are either academic mush or shallow tutorials. A free one from KDNuggets tries to hit the middle, claiming to connect the...
Nano Banana Offers Billion Indians a Shortcut to Fraud, Says Mohit Pandey
The promise of a shortcut is seductive. For a billion Indians, the Nano Banana claims to offer exactly that, a faster lane, a simpler route.
Cecilia Heyes labels language a 'cognitive gadget' for precise social learning
Language is not the engine of thought. It is a tool, remarkably precise, yet fundamentally optional to the mind that wields it.
AIM and Snowflake Host Webinar on Women Leading AI, Equity and Talent
The future of AI won’t be built on code alone, it will be shaped by who gets to lead.
Amazon adds Leo satellite internet, pledges USD 50 bn for AI HPC, 1.3 GW compute
Amazon is pulling a double move that reshapes the landscape of secure computing and connectivity.
Jensen Huang notes China employs half of global AI researchers
Jensen Huang didn’t hedge. He didn’t qualify. He simply stated a fact that should make Silicon Valley pause: half of the world’s AI researchers are...
DOE orders cloud, labs, and network integration for AI Genesis mission in 90 days
President Trump wants a new science engine built not in years, but in months. A new executive order gives the Department of Energy just 90 days to...
Meta Eyes Google TPUs, Up to One Million Units, as NVIDIA Alternative
Mark Zuckerberg is about to spend a breathtaking sum of money. Meta's capital expenditure is projected to blast past $100 billion in 2026.
CrowdStrike's Stein finds DeepSeek-R1 adds 50% more bugs on Chinese prompts
A hidden flaw in artificial intelligence isn’t a bug, it’s a feature engineered by design.
OpenAI launches ChatGPT shopping agent that uses memory data to compare products
ChatGPT Plus costs twenty dollars a month. That subscription was supposed to be the whole transaction, a clean break from the surveillance economy...
Abacus AI Review: Full-Stack Apps with Stripe, Live Dashboards, Marketplaces
Software development remains a slow, expensive grind. Abacus AI is betting it can shrink that process from weeks to minutes.
Indie developers market games as ‘AI-free,’ pledging quality without assistance
The gaming industry is hurtling toward an AI-driven future, with giants like EA, Microsoft, and Ubisoft betting big on generative tools to speed...
Microsoft's Fara-7B AI agent, rival to GPT-4o, runs on PC, logs 145k tasks
Forget the cloud. Forget the data center. Microsoft’s Fara-7B is an AI agent that lives on your PC, and it just logged 145,000 successful tasks.
Authors retract brain-mapping paper after reviewers flag fabricated citations
A researcher submits a paper on brain mapping. The reviewer’s report comes back: reject, because the citations are nonsense.
Avoid the AI-first trap as rivals tout 40% efficiency gains for your firm
When a rival promises 40% efficiency gains from AI, the reaction inside a company can be immediate.
Jony Ive and Sam Altman unveil AI hardware prototype; OpenAI plans launch under two years
The screen is dead. Long live the object. Jony Ive and Sam Altman have a prototype, a piece of AI hardware that fits in your palm, has no display,...
Gemini 3 Pro shows clear lead in coding, matching and creative writing
The old AI competition had a certain civility. That era is over. Google’s Gemini 3 Pro just landed a knockout blow in a street fight.
Amazon Deploys AI Agents to Generate Offensive Techniques and Suggest Fixes
Most security software just watches for trouble. Amazon built something that actively picks a fight with itself.
Developers flock to Docker image that removes deep-learning install lag
Ask any machine learning engineer about the worst part of the job, and they won't say training the model.
Qure.ai launches AI tools qXR and qER to spot TB, lung nodules, hemorrhages
Qure.ai is launching two AI tools to parse medical scans faster. Its qXR software scrutinizes chest X-rays for tuberculosis and lung nodules.
Developers Clash Over Google’s Cursor Killer Antigravity IDE Amid Hype
The coffee machine at every Y Combinator startup is now forking Google’s Antigravity IDE.
OpenAI research lead on ChatGPT mental-health work departs amid policy push
ChatGPT is learning to handle the most vulnerable moments in a person’s life, and doing so at a scale that feels both staggering and fragile.
TCS launches AI platform for trial oversight as India's AI use rises to 86% from 65%
Eighty-six percent. That’s the share of India's healthcare outsourcing sector now using AI, a hard leap from 65% just five years ago.
CrewAI Introduces Function-Based Guardrails for Rule-Based Output Constraints
Every tool for controlling an AI model's output is either a straitjacket or a suggestion. CrewAI now offers both at the same time.
Google DeepMind’s Nobel-winning team powers AlphaFold, nearing 1 M EMEA users
A Nobel prize-winning team. A scientific breakthrough that redefined biology. And now, nearly one million users across EMEA alone, AlphaFold is no...
Marc Benioff says he won’t return to ChatGPT after trying Gemini 3
Marc Benioff has a new daily driver. The famously opinionated Salesforce CEO tried Google's new Gemini 3 and delivered a blunt verdict on X: "Holy...
Gemini 3.0, Claude, and Grok rank GPT-5.1 top in Karpathy’s LLM Council
Silicon Valley loves a bake-off, a cage match, some ranking where one model can be declared winner.