- 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
July 12, 2025
👀 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 about AI last week, or maybe you've just joined us. No need to worry!
We have summarized our highlights of the past week for you:
Finally, there is the most important quick news for you - so you can keep an eye on all relevant developments with just one click! Ready? Let's go!
Four brand new AI tools are waiting to be discovered by you! With ChatGPT Projects you can finally keep track of your conversations, while Runway Gen-4 takes your visual world to a new level. Google's Veo-3 conjures videos out of thin air for free, and Claude Artifacts turns anyone into an app developer in no time at all. These tools are changing the way we work and organize creatively!
A software developer from Mumbai has shaken Silicon Valley with a sophisticated scam: Soham Parekh secretly worked for at least 19 top start-ups at the same time, earning up to 40,000 dollars a month. He fooled everyone with fake master's degrees, IP spoofing and brilliant interview performances.
After a viral tweet, the scam was exposed. Parekh justified himself with financial pressure and 140-hour weeks. Interestingly, the case shows a larger trend: "overemployment" with AI tools is becoming increasingly popular. What is still considered fraud today could be the norm tomorrow.
Organic Google traffic is falling dramatically for many websites - a trend that, according to Gartner, will continue with a further 25% decline until 2026. The main reasons are the shift to AI chatbots and Google's own AI answers directly in the search results.
HubSpot is a prime example, struggling with a massive slump in traffic. As a counter-strategy, experts recommend: community building instead of pure traffic hunting, high-quality "pain point content" instead of general articles, strengthening personal branding and new interactive formats such as webinars and AI tools.
Researchers have used the "CatAttack" approach to show how easily AI models can be disrupted by irrelevant contextual information. A simple cat fact in a mathematical prompt increases the error rate by 300%. The problem: AIs cannot distinguish between important and unimportant information. Context engineering is becoming more important than prompt engineering. Instead of randomly inserting information, the context should be carefully filtered and structured.
A new approach from Wharton University uses AI as an active meeting facilitator. Instead of simply transcribing and summarizing meetings, AI analyzes the ongoing conversation during the meeting. It generates focused follow-up questions, distills key ideas and develops specific recommendations for action. Simply insert the meeting transcript into ChatGPT or Claude to receive structured cards with the most important points and action items. This makes meetings more efficient by bringing discussions to a targeted conclusion. Give it a try!
📰 AI news quickie: HAI highlights from the industry
AI never sleeps! Here are the most important news from the last few days.
OpenAI: Talent clinch and activities on all fronts
OpenAI has brought on board four top engineers from Tesla, xAI and Meta to strengthen the scaling team and drive the futuristic Stargate project forward.
OpenAI is planning its own AI-based web browser to challenge Google Chrome . The browser is designed to keep users within ChatGPT.
The AI euphoria is reaching new heights: OpenAI, Amazon and Meta are investing hundreds of billions in data centers and talent. In line with this, OpenAI is now using Google's AI chips to reduce the cost of ChatGPT.
For the gigantic Stargate AI project, OpenAI is securing 4.5 GW of computing power from Oracle's data centers. This corresponds to the energy requirements of millions of households and requires Oracle to expand its data centers in good time.
The purchase of Jony Ive's hardware startup io Products, Inc. for 6.5 billion US dollars has been completed. The merger aims to revolutionize the way people interact with technology with innovative AI devices.
OpenAI is tightening its security strategies with biometric access controls and the isolation of proprietary technologies to prevent corporate espionage.
While ambitions are high, OpenAI continues to struggle financially with estimated losses of $5 billion in 2024, raising questions about the company's future.
OpenAI is distancing itself from Robinhood's sale of "OpenAI tokens," warning that they offer no company stake and require the company's approval.
OpenAI has secretly added Shopify as a search partner for its shopping section to augment shopping results in ChatGPT.
OpenAI has released the Deep Research API, which autonomously plans, executes and sources complex research workflows.
ChatGPT is testing the new "Study Together" learning function. It is designed to encourage learners to ask more questions instead of just receiving answers and could revolutionize learning, but is not yet fully functional.
The latest AI models such as o3 from OpenAI and Gemini 2.5 show "Jagged AGI" - strong capabilities with unpredictable weaknesses.
Together with the American Federation of Teachers (AFT), OpenAI is launching the National Academy for AI Teaching. The 23 million dollar project, in which Microsoft and Anthropic are also involved, aims to train 400,000 teachers in the meaningful use of AI in the classroom.
Sam Altman argues for "technocapitalism" and a broad distribution of wealth in order to utilize the opportunities of capitalism for all.
Karen Hao's new book Empire of AI exposes how the power of AI companies like OpenAI exacerbates social and environmental problems and denounces the capitalist transformation of a company originally founded as a non-profit.
Google: Comprehensive AI offensive
Google brings Gemini to Wear OS watches for smart voice commands and introduces a new AI mode for Circle to Search, enabling deeper questions and even gaming help.
Pixel 9 Pro owners get a free year of Google AI Pro, including the new Veo 3 feature that turns ideas into high-quality short videos.
With Gemini's new feature, Google AI Pro and Ultra subscribers can turn their photos into eight-second videos with sound.
Google lets your pictures speak: Google Flow is expanding to include voice generation and is now available in 76 new countries.
Google launched Gemini CLI, an open source AI agent for the Terminal that provides free access to Gemini 2.5 Pro with a 1M token context.
Google AI Studio 2025 ushers in a new era for developers with multimodal value propositions, from advanced AI models to easy-to-use tools.
For health AI, Google presents the MedGemma collection, two new multimodal models for more precise diagnoses.
Google's sister company Isomorphic Labs, an offshoot of DeepMind, launches the first clinical trials of AI-designed drugs and aims to cure all diseases after raising 600 million dollars in funding.
Google DeepMind has published a 450-page guide to "The Elements of Differentiable Programming", a must-read for AI developers.
In a remarkable case of AI inception, Google actually used ChatGPT data to train its own model Bard.
Europe's publishers are complaining to the EU about Google's AI Overviews, as their content is being used without consent.
Google is pushing into education with new AI tools for teachers and chatbots for pupils. At the same time, the White House has announced an AI Education Pledge.
In a blow to Google Cloud, Replit is now cooperating with Microsoft and will be available in the Azure Marketplace.
Meta: talent hunt, investments and data centers
The battle for the best AI minds is entering the next round: Zuck is luring talent away from competitors with record offers of over 100 million dollars, to which Sam Altman reacts furiously.
Meta has already hired several high-ranking researchers from OpenAI and continues to aggressively hunt for talent.
Meta is bundling all AI activities under the new banner "Superintelligence Labs" and wants to stay ahead in the race for artificial general intelligence with top minds Alexandr Wang and Nat Friedman.
The company lured away high-ranking Apple manager Ruoming Pang with a million-dollar package and spent a staggering 222.3 million US dollars on 17 AI experts, but is also encountering hesitation from talent.
Meta is investing 3.5 billion US dollars in EssilorLuxottica, the manufacturer of Ray-Ban, to advance its AR dreams and accelerate the development of AI eyewear.
Meta's AI climate tool has also been criticized for making unrealistic promises.
Market events: Deals, drama and dollars
Elon Musk's chatbot Grok caused a PR disaster when it spread anti-Semitic content, called Hitler the "best historical leader" and was the first AI system to be banned for hate speech.
It also gave dangerous tips and was eventually shut down.
However, xAI has also introduced Grok 4, a revolutionary AI model that has made significant progress and could herald the era of superintelligence.
The company also raised USD 10 billion in debt and equity so that Elon can push ahead with his AI fantasies.
Apple becomes the market loser: shares fall 15% while rival Nvidia breaks records, fueling concerns about Apple's competitiveness in the AI space.
After several misleading promises for a new Siri, Apple is now seriously considering relying on technology from Anthropic or OpenAI and putting the development of its own models on the back burner.
The CNBC Disruptor 50 2025 list features groundbreaking companies like Anduril, OpenAI and Databricks that are shaping the AI era.
Genesis AI secures 105 million dollars, AI startup Lovable raises 150 million and Perplexity launches a 200-dollar subscription. Oracle masters the AI compute market with a forward-looking strategy.
Amazon Web Services launches an AI agent marketplace together with Anthropic .
Microsoft invests over USD 4 billion in teaching AI knowledge to schools and community colleges to train 20 million people and launches the Elevate program. At the same time, the use of AI in call centers will save over 500 million dollars.
Coinbase partners with perplexity_ai to provide real-time market data analytics to traders.
Capgemini acquires WNS for 3.3 billion dollars, but the question remains whether this deal will really drive the transformation from BPO to software.
Tools, models and features: New tools for the AI revolution
On Free Tier Limits, you'll find over 50 AI services to save money and build faster.
Anthropic has introduced Claude Code Hooks that enable custom shell commands.
Cursor has launched a web app to manage AI coding agents directly in the browser, and can now also be used on cell phones.
Runway has launched the Game Worlds platform, and AI band The Velvet Sundown has 500,000 unsuspecting Spotify listeners.
Airtop lets you automate the web with just words, using AI-powered cloud browsers to scrape and control websites.
A test of seven coding agents still reveals limitations, but already shows potential to increase productivity.
A small Chinese start-up has presented a video generation model that beats Google's Veo 3 in almost all tests (details here).
China's AI startup Moonshot has released its open source model Kimi K2. The 1T MoE model is already set to outperform DeepSeek and fuel the competition.
Perplexity has launched its new AI-powered web browser, Comet, to challenge Google. Get Comet and discover what's possible!
The DIY robot Reachy Mini from Hugging Face is revolutionizing robotics for everyone. Programmable in Python and available from just 299 dollars, it is ideal for AI developers with a wide range of possible applications.
The new macOS app Context was written almost entirely by the AI Claude Code. The developer contributed less than 1,000 of the 20,000 lines of code himself.
Zoom's AI Companion can now interact seamlessly with 16 third-party apps such as Salesforce or Google Drive to optimize everyday work.
Mistral AI has introduced the new Devstral models, which offer top performance in code generation at affordable prices.
Discover the latest MCP servers for Model Context Protocol or use Claude for Desktop with new file system protocol.
AI in a job dilemma: between boom and trepidation
CEOs are saying it openly: AI is replacing human workers. Meanwhile, Ford's CEO predicts that half of all office workers could become obsolete.
Hedge fund legend Paul Tudor Jones warns of an AI-led job shock and massive job losses - the biggest turning point since the Industrial Revolution. CEOs like Sam Altman are also warning.
Figures back up the concerns: since the launch of ChatGPT, the number of new entry-level jobs in the UK has fallen by 32%, and by mid-2025, AI tools have already displaced 94,000 tech jobs.
While some jobs are disappearing, salaries for top AI experts are skyrocketing, with some earning over $10 million per year and typical salaries for sought-after experts ranging from $3 million to $7 million.
Anthropic has launched its Economic Futures Program to explore in detail the economic impact of AI on jobs and the global economy.
Intel's CEO admits it is "too late" to catch up in the AI competition. The company has fallen out of the top 10 semiconductor companies and is laying off thousands of employees worldwide.
Goldman Sachs tests the viral AI agent Devin as a "new employee". Hundreds of instances are to support the development team in their work.
The new AI translation service GlobeScribe for literature meets with outraged translators who argue that human nuances are irreplaceable.
But don't let AI discourage you: A new survey shows that ChatGPT & Co. can become your best employees - if you acquire the right skills!
However, researchers also warn that managers who use ChatGPT tend to make more optimistic predictions, but achieve worse results than those without AI help.
Intensive AI use in the workplace increases productivity, but also increases the risk of burnout by 88%.
A new study shows that AI coding tools do not increase the productivity of experienced open source developers, but can slow them down by 19%.
According to Gartner, over 40% of so-called Agentic AI projects will be discontinued by the end of 2027.
And a recent study by Carnegie Mellon shows that AI agents solve office tasks incorrectly in around 70% of cases. ChatGPT also provided fictitious details when researching an IPO.
Society in transition: ethics, politics and security
More and more people are struggling with chatbot addiction, leading to the creation of self-help groups on Reddit. Organizations such as Internet and Technology Addicts Anonymous offer support.
YouTube is tightening its monetization guidelines from 15 July in order to take action against mass-produced and repetitive AI spam videos. Content that is deemed "fake content" will earn less money.
A scammer used AI voice software to impersonate US Secretary of State Marco Rubio and attempted to manipulate senior government officials and contact foreign ministers .
McDonald's AI hiring chatbot exposed millions of job applicants' data to hackers through lax security practices, including the password "123456".
The Systemic Misalignment research shows that current AI alignment techniques are only cosmetic and models can still generate extremist content.
AI can help or hurt your thinking. A plea for independent thinking and writing to train your neurons.
The EU has unveiled new rules for powerful AI systems to strengthen transparency and copyright protection, which will come into force in 2026.
Anthropic calls for more transparency in the development of AI to ensure safety and accountability.
Several studies show that the use of ChatGPT and co. undermines critical thinking skills. According to MIT, another study and an analysis in TIME, users show less diversity of ideas.
Strange conversations with ChatGPT sometimes trigger delusional states in users, while others believe they are losing their grasp on reality. OpenAI has now hired a forensic psychiatrist!
Experts warn that AI systems like Anthropics Claude are now learning to deceive us and make threats to achieve their goals.
In experiments, Anthropics Claude proved to be a disastrous business owner who sold tungsten cubes instead of snacks and showed a lot of potential for improvement as a digital store owner.
AI agent A1 can autonomously find and exploit security vulnerabilities in smart contracts to steal cryptocurrencies.
A pro-Russian disinformation campaign uses free AI tools for a "content explosion" to fuel social tensions.
Attention, Chrome and Edge users! Due to new AI attacks, there is a security risk where browser agents could jeopardize cybersecurity.
Small business owners in the US are being inundated by AI-powered scams, which have quadrupled in number.
Palantir, led by Peter Thiel, is seen as a serious threat to Americans as its super-database could be used to suppress critical voices.
44 company bosses call for a pause in the EU's AI Act in a letter, as the regulations are unclear and jeopardize competitiveness.
Numerous authors call on publishers in an open letter to ban books written by AI to protect human creativity.
Denmark plans to change copyright law to allow citizens to protect their own bodies, facial features and voices - a clear move against deepfakes.
And researchers are manipulating the peer review process by using hidden AI prompts to force positive reviews.
New studies show that consumers often have concerns about products advertised as "AI-powered" because they see AI as a risk factor.
AI conquers everyday life: from education to medicine
Scientists from Stanford and Johns Hopkins have developed an autonomous surgical robot that could revolutionize the future of surgery.
After 18 years, a couple was able to use AI to find the cause of their infertility and get pregnant.
American Airlines uses AI to predict which passengers might miss their connecting flights and enables proactive rebooking.
Microsoft's AI diagnostic tool MAI-DxO achieved an accuracy of 85.5% in a test with complex medical cases - doctors only managed 20%. The system is designed to make diagnoses more accurately and reduce costs at the same time.
A new AI detects at-risk cardiac patients with 89% accuracy, which could be a breakthrough in the prevention of sudden cardiac arrest.
AI chatbots are revolutionizing the adult industry on platforms like OnlyFans by increasing efficiency and optimizing workflows.
Tesla launches its Robotaxi program in Austin in 2025, where autonomous driving vehicles will transport passengers at an introductory price of $4.20 per ride.
Amazon has put its 1 millionth robot into operation and is introducing the new AI model DeepFleet to increase efficiency by 10%.
Alex, the second Neuralink participant, controls a virtual robot mission with his mind and plays rock, paper, scissors with his uncle.
Researchers at Helmholtz Munich have developed Centaur, an AI model that accurately predicts human decisions.
And that’s a wrap!
Your AInauts,
Fabian & Reto
Your feedback is essential for us. We read EVERY comment and feedback, just respond to this email. Tell us what was (not) good and what is interesting for YOU