👀 Weekly AI News: Did You Miss It?

👨‍🚀 The most important AI updates at a glance

Author

Die AInauten
October 25, 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!

Become An AI Expert In Just 5 Minutes

If you’re a decision maker at your company, you need to be on the bleeding edge of, well, everything. But before you go signing up for seminars, conferences, lunch ‘n learns, and all that jazz, just know there’s a far better (and simpler) way: Subscribing to The Deep View.

This daily newsletter condenses everything you need to know about the latest and greatest AI developments into a 5-minute read. Squeeze it into your morning coffee break and before you know it, you’ll be an expert too.

Subscribe right here. It’s totally free, wildly informative, and trusted by 600,000+ readers at Google, Meta, Microsoft, and beyond.

Anthropic introduces "Skills," an exciting new feature for Claude that functions like a specialized toolkit. These modular power-ups transform your AI assistant from a generalist to a specialist by providing domain-specific knowledge, workflows, and code as needed.

The possible applications are diverse: from uniform branding and automated reports to efficient customer support. Skills elegantly solve the problem of repeated context provisioning through structured, on-demand expert knowledge—usable without any developer knowledge. Be sure to give it a try!

🫶 Favorite cartoon characters at the touch of a button

With Gemini, you can create magical cartoon worlds! Transform your photos into Pixar-like scenes featuring your favorite characters—from Super Mario to Mickey Mouse, everyone's included! The workflow is super simple: launch Gemini, upload your image, customize the prompt, and you're good to go. The trick is in the detail prompt: photorealistic 3D scenes, dramatic lighting, and crystal-clear textures make all the difference!

OpenAI has introduced its own web browser, ChatGPT Atlas! The browser combines classic search functions with AI-supported features such as context-sensitive chats, a handy ChatGPT bubble at the cursor, and an agent mode for autonomous browsing. Particularly exciting is the integration of the memory function, which combines chat histories, custom instructions, and browser history.

While similar features are already offered by competitors such as Comet, the comprehensive context provided by the memory function could create real added value. However, potential security risks from prompt injection attacks should be noted, especially in agent mode...

Be careful when copying and pasting in AI tools: customer data and sensitive information end up directly on external servers belonging to OpenAI & Co. The 3-second rule helps: if you wouldn't share it in a WhatsApp group, it doesn't belong in the chat either.

Alternatives include: deactivating Memory & Training, anonymizing data, or using enterprise versions. For companies, there will be a HubSpot webinar on November 10 on the legally compliant use of AI. Save the date!

App builder Lovable is shaking up e-commerce with its new Shopify integration! Now users can create a complete online store simply by chatting.

This is especially great for non-techies, as it further lowers Shopify's already low barrier to entry. It allows virtually anyone to start their own online business—without any technical know-how.

📰 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 on course for expansion: browsers, erotica, and global plans

Anthropic & Google fight back: More affordable models and smart agents

  • Anthropic drops Claude Haiku 4.5 and promises Sonnet 4-level coding power for only a third of the cost at twice the speed. The API is available from $1/$5 per million tokens and, according to Ars Technica, is ideal for real-time chats, copilots, and agent orchestration.

  • Claude gains skills: You can create modular folders with instructions, resources, and code that Claude automatically loads for apps, code, and APIs, reducing prompt tinkering.

  • Google launches Gemini Agent and Gemini Computer Use: These AI agents act as web autopilots, clicking, typing, etc. for you.

  • Google is pimping up AI Studio and announcing the "Vibe Coding Experience" with one-click deploy, an app gallery, and modular "superpowers" to make developing AI apps smoother than ever before.

  • Anthropic's revenue targets are ambitious: by 2026, annualized recurring revenue is expected to nearly triple to $26 billion. Growth will be driven by strong enterprise demand and the favorable Haiku model.

  • Claude now connects to Microsoft 365 and enables enterprise search via SharePoint, OneDrive, Outlook, and Teams, eliminating the need for context gathering.

  • Claude gets a long-term memory: For Pro/Max, the bot saves your preferences, which can be edited separately for each project. And with the desktop app, you can add Claude to any application on macOS and Windows via a shortcut.

  • In its Earth AI features and updates, Google links Gemini with geographic, weather, and population data to analyze global risks such as storms or algal blooms.

The infrastructure spending spree: Tech giants invest trillions

  • OpenAI is forging a five-year plan to finance pledged expenditures of over $1 trillion. The plan includes new revenue streams, debt deals, and fresh capital to fuel the turbocharged rush toward AI infrastructure and new products.

  • In the largest data center deal ever, BlackRock, Microsoft, Nvidia, and xAI acquire Aligned Data Centers for $40 billion. The first move by the AI Infrastructure Partnership (AIP) secures over 80 data centers with 5 GW capacity – Compute is the new oil.

  • To increase its AI power, Microsoft is ordering 200,000 Nvidia chips with a potential value of $14 billion through an Nscale deal. The first GPUs will arrive next year from Texas/Portugal, with another 52,000 coming from Norway.

  • Meta begins construction of a data center in El Paso with a capacity of up to 1 GW. With an initial investment of US$1.5 billion, 100% renewable energy, and 200% water restoration, the infrastructure for superintelligence is to be delivered—creating around 1,800 construction and 100 operational jobs.

  • OpenAI is forging an alliance with Broadcom to develop its own inference chips. From the end of next year, 10 GW of computing power will be available via Broadcom Ethernet to reduce dependence on Nvidia/AMD and speed up the expansion of the company's own data centers.

  • Intel is making a comeback with its AI GPU 'Crescent Island', which will be released next year and is optimized for energy efficiency in inference tasks. An interesting detail: according to Reuters, Nvidia is investing $5 billion in its competitor.

  • Amazon secures energy supply for its data centers and provides initial insights into its "Cascade" nuclear reactor project. The small modular reactor is expected to deliver 960 MW and go into construction at the end of the decade.

Market upheaval: AI hype, bubbles, and new centers of power

  • The US is riding the AI hype train – if the bubble bursts, there could be massive turmoil on the stock market and job market. Should the bubble be burst to reduce lock-in effects and promote genuine use cases?

  • IMF Chief Economist Pierre-Olivier Gourinchas warns of a bubble risk in the AI boom reminiscent of 1999. However, with little debt involved and investment below 0.4% of GDP, a system crash is unlikely.

  • Investors analyze the risks in AI trading: High S&P 500 exposure, hyperscalers' capex spending rising to $500 billion, power shortages, and "circular" deals could put pressure on margins.

  • Bye, FAANG: We are now living in the MANGO era! Microsoft, Anthropic, Nvidia, Google DeepMind, OpenAI. Power is shifting to the AI leaders. The result: fewer entry-level jobs and a battle for top researchers.

  • Nvidia suffers a massive market share crash from 95% to 0% in China after CEO Jensen Huang called US export rules a mistake—while China (including Huawei) develops its own AI chips.

  • The OpenAI effect: Partnerships boost shares in NVIDIA, Broadcom, and Oracle, while others come under pressure.

New tools for developers, creatives, and everyday life

  • Google pimps the Gemini CLI for developers: Tools such as vim, htop, and git now run interactively in context.

  • The Claude Code Plugins Marketplace lists 227 production-ready plugins ranging from security audits to dev tools, which you can use to build your AI coding sidekick in a modular way.

  • Bubble launches the AI Agent, which allows you to build apps using drag-and-drop or chat.

  • Director automates web tasks via prompts and scales with up to 25 parallel browsers and 1Password integration.

  • ChatGPT's new Spotify app integration becomes your personal AI DJ, analyzing your listening history, recommending suitable bands, and creating playlists on demand. Spotify itself is expanding its AI DJ for premium users.

  • With Shortcut, AI comes natively to Excel to create models in minutes that remain formula-driven and auditable.

  • Facebook lets Meta AI search through your pictures to suggest the best shots and create collages—all privately until you post (opt-in). Instagram gets a similar feature for Stories with Meta AI "Restyle".

  • Microsoft is also giving Paint a restyle AI upgrade to restyle images using style presets (currently only available for Windows Insiders on Snapdragon PCs).

  • YouTube rolls out AI-powered Q&A stickers, daily shorts time limits, and desktop communities.

Society, security, and the human factor

  • OpenAI's Sora 2 generates deceptively real misinformation in video form—watermarks can be easily removed.

  • Following AI deepfakes, OpenAI strengthens its guardrails, pauses the generation of MLK videos, and apologizes.

  • A new, measurable AGI definition classifies the capabilities of AI: GPT-4 achieves 27% of the level of an educated adult, an early version of GPT-5 already 57%—long-term memory is still its greatest weakness.

  • A fierce legal battle is brewing: Reddit is suing Perplexity because the AI search engine is allegedly scraping Reddit content instead of paying for it.

  • Top managers warn: A "human-in-the-loop" approach to AI is not enough; crystal-clear judgment is needed. Managers are becoming "chief experimentation officers" because judgment is the new prompting.

  • AI-powered rent software is being banned in New York because its use is considered illegal price fixing.

  • AI luminaries such as Hinton and Bengio are calling for a ban on superintelligence until safety and societal approval have been clarified.

  • Meta is cutting 600 AI jobs at its Superintelligence Lab in order to reduce coordination and increase impact, according to AI chief Alexandr Wang.

  • Amazon aims to save over 600,000 jobs by 2033 through AI robots and automate 75% of its processes.

That’s a wrap for now! See you next week,
Fabian & Reto

Follow us on Twitter & LinkedIn!

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!