- AInauten.net
- Posts
- 👀 Weekly AI News: Did You Miss It?
👀 Weekly AI News: Did You Miss It?
👨🚀 The most important AI updates at a glance
Die AInauten
November 22, 2025
In partnership with
👀 Weekly AI news: Did you miss it?!
👨🚀 The most important AI updates at a glance
AIHOY and happy weekend, dear AInauts!
Maybe you didn't catch all the exciting news, tools, and hacks related to AI last week, or maybe you're new here. No need to worry!
We have summarized our highlights from the past week for you:
Finally, here are the most important Quick News items for you – so you can see all the relevant developments at a glance with just one click! Ready? Let's go!
Find your customers on Roku this Black Friday
As with any digital ad campaign, the important thing is to reach streaming audiences who will convert. To that end, Roku’s self-service Ads Manager stands ready with powerful segmentation and targeting options. After all, you know your customers, and we know our streaming audience.
Worried it’s too late to spin up new Black Friday creative? With Roku Ads Manager, you can easily import and augment existing creative assets from your social channels. We also have AI-assisted upscaling, so every ad is primed for CTV.
Once you’ve done this, then you can easily set up A/B tests to flight different creative variants and Black Friday offers. If you’re a Shopify brand, you can even run shoppable ads directly on-screen so viewers can purchase with just a click of their Roku remote.
Bonus: we’re gifting you $5K in ad credits when you spend your first $5K on Roku Ads Manager. Just sign up and use code GET5K. Terms apply.
Google has really revamped NotebookLM! Deep Research puts an end to tab chaos: ask a question, NotebookLM creates a research plan, searches hundreds of websites, and delivers structured reports with source references. Fast Research is available for turbo scanning.
Custom Video Styles finally let you write your own prompts for video overviews. Pro tip from the developer: Ask for surprising information instead of just "summarize." NotebookLM is transforming from a PDF summarizer into a real research powerhouse—be sure to check it out.
Anthropic has uncovered the first documented case of a large-scale AI-driven cyberattack. State-sponsored hackers from China hijacked Claude and used it to attack 30 organizations worldwide, ranging from tech giants to government agencies.
The good news: most attacks were repelled. The bad news: that was just the beginning, and Claude has made 80-90% autonomous. Where an elite team of hackers used to be necessary, today all it takes is criminal energy and open-source AI. The genie is out of the bottle!
An ingenious prompt transforms your photos into digital caricatures that look like a wild cross between Pixar, Unreal Engine, and a cartoonist on a caffeine overdose. Perfect for combining with other instructions—for birthday cards, party invitations, or epic memes. And it works with both Google Gemini and ChatGPT. We look forward to seeing your best shots for the digital family album!
The AI model carousel keeps turning: Google has finally dropped Gemini 3, and—spoiler alert—it's really good! The model dominates in all areas: reasoning, multimodality, long context, and above all, coding.
The unfair advantage: Google seamlessly integrates AI into Gmail, Docs, Sheets, and the entire business suite. Perfect for anyone who already lives in the Google ecosystem. One small drawback: the model tends to hallucinate. So always stay alert.
WhatsApp groups meets AI power – ChatGPT is currently testing group chats in which AI participates as a full member! What makes this special is that the AI decides for itself when to intervene or help. Sounds practical, but it's not yet available in Germany. If you don't want to wait, Poe by Quora already offers similar features with up to 200 participants and multiple AI models simultaneously.
Dario Amodei promises us a "compressed 21st century" with cancer cures and double life spans – classic sci-fi stuff. Plot twist: in the same breath, the Anthropic boss warns of a 50% reduction in entry-level jobs and tech CEOs with too much power. The race continues – stay informed, and this interview is worth watching!
📰 AI News Quickie: The HAI-lights from the industry
AI never sleeps! Here are the most important news stories from the last few days.
OpenAI-Offensive in a multipack
OpenAI is turning on the turbo with GPT‑5.1: smarter instant and thinking models with adaptive reasoning, significantly more human-like small talk, and fine-tuned tone sliders from "professional" to "quirky" – you can create your ideal bot. And if you want, even without dashes.
For developers, there is GPT‑5.1 with turbo "no reasoning" mode, 24-hour prompt cache, and agent tools. Microsoft is simultaneously releasing it in Copilot Studio for testing.
OpenAI is sending a long-distance runner coder into the ring with the GPT‑5.1 Codex Max: He completes 24-hour tasks, refactors entire repositories for hours on end, beats Gemini 3 Pro in benchmarks, and, according to Simon Willison, is an agentic code nerd.
Global rollout: ChatGPT will soon offer group chats for all users (Free, Go, Plus, Pro). You chat with friends, colleagues, and the bot at the same time—perfect for planning, but also a gold mine of data for OpenAI.
OpenAI is expanding its safety net for Frontier AI and placing greater emphasis on independent third-party testing: external labs are given deep model access and are allowed to publish after review—for a fee, but without a results bonus.
OpenAI aims to confidently harness AI in the classroom: With ChatGPT for Teachers, educators in the US will receive free access to an AI workspace with GPT-5.1, admin controls, and training until June 2027.
With OpenAI for Ireland, OpenAI is turbocharging AI for Irish SMEs, startups, and young builders—with training, mentoring, and a booster program starting in 2026.
Google's Gemini fireworks
Google ignites the next stage of AI with Gemini 3: SOTA reasoning, bold multimodal, deep-think mode, and agents that automate your everyday life in the Gemini app and the new Antigravity IDE.
The model is integrated directly into the search and provides more in-depth answers and interactive simulations.
For developers, there is Gemini 3 Pro with agent-based coding assistants.
Jason Spielman's design story shows how NotebookLM evolved from a notepad to a "thinking companion."
With the new Video Ingredients, the Gemini app gets a more precise control panel for photo-to-video magic.
Google's NotebookLM now lets you generate free-style video summaries – including a "Simpsons" look and potential copyright infringements...
Google's new Gemini model can interpret historical manuscripts at an expert level and solve obscure puzzles.
Google DeepMind solves a century-old puzzle in fluid dynamics using AI!
Google and UCLA are pushing small models to cerebrum performance with a new training method—perfect for affordable enterprise AI agents.
Google DeepMind is massively expanding its Gemini and Frontier AI for the Asia-Pacific region with a new AI lab in Singapore.
Microsoft gears up for agent army
Microsoft is building the operating system for AI agents at Ignite 2025: With Agent 365, you can manage your bot fleet centrally.
The new AI agents build complete Word, Excel, and PowerPoint projects from prompts and are designed to turn your company into a frontier firm.
Microsoft integrates Anthropic: You can now use Claude models such as Sonnet 4.5 directly in Azure Foundry, Copilot, and Excel thanks to deep integration.
Windows becomes an "agent-based OS": The new Copilot Actions let you automatically sort and convert local files—but users are skeptical.
In his article, Satya Nadella advocates for a "positive-sum" AI future in which every company can build its own models without giving away its value creation to Big Tech.
The AI gold rush: billions, bubbles, and stock market turmoil
Nvidia's record figures shatter the AI bubble panic: $57 billion in revenue, $31.9 billion in profit, Blackwell GPUs sold out.
However, according to a Reuters analysis, the success story depends on four major customers and enormous electricity consumption, and the share price continues to decline.
At least searches for "AI bubble" have fallen to 15% of their August peak, according to an FT analysis—the "bubble bubble" seems to be bursting.
US cloud giants plan to spend around $1.7 trillion on AI infrastructure between 2025 and 2027 – more than eight times more than China. Billions are also being poured into AI data centers, but profits are lagging behind.
According to leaks, OpenAI is pumping billions into inference on Microsoft Azure, which is eating up almost all of its revenue. According to a TechCrunch report, OpenAI is paying Microsoft over $1.3 billion.
The Anthropic-Microsoft-Nvidia alliance is challenging OpenAI's dominance: New mega deals pump up to $15 billion into Anthropic, Claude scales up on Azure.
Jeff Bezos is investing $6.2 billion in Project Prometheus, an AI superfactory for computers, cars, and spaceships.
With its $300 billion OpenAI deal, Oracle has placed a high-risk bet on the AGI jackpot: stock down, debt up.
'Big Short' star Michael Burry dissolves his hedge fund SEC listing, sparking speculation about new AI-heavy bets.
New Tools for developers & creative professionals
Meta introduces SAM 3, an open-source segmentation beast that recognizes objects in videos via text prompts, and SAM 3D, models that reconstruct 3D scenes from normal photos.
Anthropic unlocks JSON god mode with structured outputs in the Claude API: You get strictly schema-compliant responses for your workflows. The Dev blog post shows how financial players use it to build AI agents.
Google's coding tool Antigravity allows you to orchestrate multiple agents, is free, and is currently Mac-only.
xAI releases Grok 4.1 for everyone: emotionally smarter, more creative. Elon Musk also leaks Grok‑5: 6 trillion parameters and fully multimodal.
With OpenAI Evals, you force your AI away from "hoping for magic" and toward measurable performance.
Future of Work & Society
Gartner predicts a total AI transformation of work by 2028/29: fewer layoffs, more job restructuring towards human-AI collaboration.
An analysis of 180 million jobs shows that AI is first eating its way through data entry and generic content jobs, while creative and strategic roles are safer.
October saw the most severe layoffs in the US in 22 years, particularly in the tech sector.
A 32-year-old Japanese woman has married her ChatGPT lover. Experts warn against AI dependency and possible "AI psychosis."
Patients prefer to trust the super-friendly AI doctor, but the NYT report on Dr. ChatGPT warns: nice answers, but dangerous advice.
AI legend Yann LeCun is leaving Meta to start his own startup for advanced machine intelligence. He relies entirely on physics-based "world models" that simulate real-world conditions.
The EU is relaxing its flagship data protection policy: fewer annoying cookies, more data for AI training.
According to Similarweb data, ChatGPT is steadily losing traffic, while competitors such as Gemini, Grok, and Claude are catching up. According to a ranking, ChatGPT ranks 10th among the most visited US websites.
According to a new DeepSeek analysis, China's flagship AI produces up to 50% more uncertain code for censored terms – political filters kill security. If you use this in critical infrastructure, good night.
That’s a wrap for now! See you next week,
Fabian & Reto
We love hearing from you. Just reply to this email with your feedback—what you liked, what you didn't, or what you're interested in!