- AInauten.net
- Posts
- ๐ Weekly AI News: Did You Miss It?
๐ Weekly AI News: Did You Miss It?
๐จโ๐ The most important AI updates at a glance
๐ Weekly AI news: Did you miss it?!
๐จโ๐ 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 โ so you can see all the relevant developments at a glance with just one click! Ready? Let's go!
Imagine this scenario: six leading AI models compete against each other in a spectacular trading competition. With $10,000 in starting capital each, Claude 4.5, DeepSeek V3.1, Gemini 2.5 Pro, GPT-5, Grok 4, and Qwen 3 Max trade six different cryptocurrencies completely autonomously with 3x leverage and real-time trades.
Their emotional reactions to wins and losses are particularly fascinatingโranging from self-doubt to overconfident trades. The models develop surprisingly human traits in their trading decisions. ๐จ๐ณ vs. ๐บ๐ธ โ and the winner is โฆ
Marketing tools often cause media breaks due to constant copying and pasting between different platforms. This leads to loss of data, context, and momentum. SwissMadeMarketing shows that there is another way. They have developed a solution that intelligently connects tools instead of creating new ones.
Try it out for yourself with the smart business idea validator, which analyzes your website or idea in 2 minutes, generates SEO keywords and ChatGPT questions, and keeps this as context for all further marketing activities. Really powerful!
A small but telltale detail immediately exposes AI-generated texts: the excessive use of em dashes (โ). These dashes are a typical feature of ChatGPT outputs, especially in English. To avoid this, you can add a simple instruction to the custom instructions that systematically replaces em dashes with periods or commas.
A privacy-first chatbot from Switzerland is making waves! Lumo AI, developed by data protection pioneer Proton, uses various open-source language models (mainly Mistral) and independently decides which model is best suited to each query. The bot scores points for its simple operation, text data import, and web search. Although it doesn't quite match the big players in terms of features, it impresses with sensitive queries that focus on data protection.
The 2025 Coca-Cola Christmas ad shows progress in AI video production compared to last yearโit is longer, more consistent, and produced with fewer staff. For optimal results with AI-generated videos and photos, technical terms from film language are crucial: camera angles (wide, low, OTS), lighting effects, and composition.
Google Gemini is expanding its feature set with the new Canvas function, which can automatically create Google Slides presentations. With a simple prompt, "Please build me a presentation," Gemini creates complete slides that can be edited directly. It is particularly effective when used with pre-trained assistants (Gems) and seamlessly integrated into the Google ecosystem.
๐ฐ 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: Billion-dollar deals, user explosion, and internal drama
OpenAI signs a seven-year, $38 billion cloud deal with Amazon, thereby loosening its exclusive ties to Microsoft. The partnership secures more computing power through hundreds of thousands of NVIDIA GPUs for model training and agent development.
OpenAI and Microsoft plan to build the $1.4 trillion supercomputer "Stargate" with 30 GW of power, while Meta is working on the 5-GW computing monster "Hyperion", but is struggling with funding gaps.
The company is aiming for annual sales of over $20 billion and wants to achieve triple-digit billion sales by 2030. CEO Sam Altman makes it clear that despite trillion-dollar investment plans, there will be no government guarantees; instead, the company is relying on the free market and private financing.
OpenAI reports impressive growth figures with 1 million paying corporate customers and 800 million weekly users, making it the fastest-growing business platform.
In an explosive affidavit, co-founder Ilya Sutskever describes CEO Sam Altman as a manipulative chaos operator, fueling internal drama and revealing earlier merger plans with Anthropic.
OpenAI's AI video studio Sora has launched as a TikTok-style Android app in the US, Canada, and parts of Asia, with Europe to follow soon. Users can create and remix AI videos directly on their phones and purchase additional creations.
According to new guidelines, ChatGPT no longer provides tailored health, legal, or financial advice in order to reduce the risk of lawsuits. This is a tightening of existing rules.
A new ChatGPT feature allows you to interrupt ongoing responses, add new context, and have the response updated without losing progress.
Developers get more power with GPT-5 Codex Mini, 50% higher rate limits for paying users, and prioritized processing.
With Aardvark, OpenAI is launching a GPT-5-powered AI security researcher that scans repositories, validates exploits, and suggests fixes.
A user negotiated a $195,000 bill down to $33,000 with Claude, while Kim Kardashian blames ChatGPT for failing her law exams.
OpenAI has been sued because ChatGPT allegedly encouraged suicide, which could fundamentally change the rules for AI conversations.
To protect young people, the Teen Safety Blueprint was introduced, which provides for age-appropriate design and stronger protective mechanisms.
Google: Gemini becomes a universal assistant
Gemini is being integrated more deeply into Google Workspace: The Deep Research feature analyzes data from Gmail, Drive, and Docs for comprehensive reports and is available without an additional subscription.
In Google Docs, Gemini also uses linked files in the document as a knowledge base to reduce hallucinations.
With Gemini Canvas, you can create complete Google Slides presentations, including text, images, and design, from a simple prompt.
Google Maps is getting a Gemini upgrade with voice-controlled navigation, landmark turn-by-turn directions, and proactive traffic alerts.
Google Finance is being enhanced with Gemini-powered Deep Search, which provides quoted answers, research plans, and forecast data for investors.
Google's no-code app builder Opal is launching in over 160 countries, enabling the rapid creation of mini-applications and workflows.
Google plans to test data centers in space starting in 2027 with its Project Suncatcher. NVIDIA's H100 GPU is already being tested in space.
Anthropic: European expansion and deeper insights
Anthropic is expanding into Europe and opening offices in Munich and Paris to meet strong demand from customers such as BMW and SAP.
Researchers have discovered that Claude can perceive its own internal states in approximately 20% of cases and can even rationalize forced responses, which presents opportunities for audits but also new risks.
In Iceland, Anthropic is launching a national AI education pilot program, providing teachers across the country with access to and training for Claude.
The Economic Futures Programme in the UK/EU aims to use data and grants to put political decisions in the field of AI on a sound footing.
Microsoft: Offensive with super AI and new agents
Microsoft AI introduces the Humanist Superintelligence (HSI) initiative, a human-centered super AI with a focus on medicine, AI companions, and energy.
A new agent in Microsoft 365 Copilot called Researcher with Computer Use can independently search the internet and operate apps.
Microsoft's new image model MAI-Image-1 has risen to the top 10 on LMArena and is already being used in Bing Image Creator, but is not yet available in the EU.
Microsoft and Lambda have signed a multi-year, billion-dollar deal to build an AI infrastructure with tens of thousands of NVIDIA GPUs and increase the availability of computing power.
Market & Investments: Between Trillion-Dollar Bets and Crash Warnings
Concerns about an AI bubble are growing: stock prices are falling, and Michael "Big Short" Burry is betting ~$1.1 billion against Nvidia and Palantir.
According to Bloomberg, Apple plans to pay Google $1 billion annually to use the 1.2 trillion parameter model Gemini for a new version of Siri.
NVIDIA CEO Jensen Huang and Chief Scientist Bill Dally were awarded the Queen Elizabeth Prize for Engineering for their pioneering work in GPU technology.
NVIDIA is collaborating with Samsung and the Hyundai Motor Group to build "AI Factories" to accelerate smart manufacturing and vehicle development.
Der deutsche KI-Pionier Aleph Alpha strukturiert um: Grรผnder Jonas Andrulis wechselt in den Beirat, der Fokus liegt nun auf Behรถrden- und Enterprise-KI.
Waymo erweitert seinen Robotaxi-Service 2026 auf Las Vegas, San Diego und Detroit.
New tools, assistants, and agents
Ethan Mollick's guide provides practical tips on how to make the best use of different AI models for different tasks.
Perplexity is testing a virtual fitting avatar for clothing and has signed a licensing deal with Getty Images.
Meta is introducing Vibes in the Meta AI app, a TikTok-like AI video feed, in Europe.
Canva has released its own design model that provides editable layouts and integrates AI features throughout the platform.
Adobe Photoshop and Express will be integrated directly into ChatGPT, enabling the creation of designs via chat.
DeepL is expanding with an autonomous AI agent that can operate enterprise applications and automate processes such as sales research.
ClickUp 4.0 launches with two AI agents and a new search function, launching an attack on Slack and Notion.
Former Meta designers are launching the Stream Ring, a "mouse for the voice" that allows users to take notes by whispering.
AI in practice: Law, ethics, and society
Collins Dictionary has named "vibe coding", a term coined by Andrej Karpathy, as its word of the year for 2025: programming based on intuition rather than strict rules.
Universal Music Group signs strategic deals with Udio and Stability AI for licensed AI music tools, while Japanese studios such as Ghibli are taking legal action against Sora.
The OpenAI Assistants API was secretly misused by hackers as a C2 channel for espionage, while xAI forced employees to provide biometric data for an AI girlfriend bot.
Research by The Atlantic shows that the data crawler Common Crawl gives AI companies access to articles behind paywalls.
Thatโs a wrap for now! See you next week,
Fabian & Reto
We love hearing from you. Just reply to this email with your feedbackโwhat you liked, what you didn't, or what you're interested in!
