Weekly AI News

Hey it’s Jul,
Greetings and welcome to the twenty-first edition of “Weekly AI News”!

Last week, Sam Altman played Santa with o3-Pro and its 80% price cuts - enough to put summer sales to shame. And the brilliant thing about o3-Pro is that it becomes genius-level when you feed it lots of context. A bit like that friend who needs three coffees before saying something intelligent.

On Apple's side, WWDC was... how should I put it... cautious? No Siri revolution, but they're finally opening their models to developers. That's their strategy: make AI invisible but indispensable. We'll see if they pull it off...

Meanwhile, everyone wants to control EVERYTHING. Nvidia is forcing cloud startups to go through its marketplace, Mistral is launching its own cloud, AMD is hiring to do more than just chips... It's a race to see who'll control the entire stack.

Bonus: Disney is suing Midjourney for unauthorized Yodas. Because come on, let's not get crazy here! 😄

Happy reading!

🍎🤖 Apple Unveils AI Upgrades at WWDC 2025

Apple announced significant AI integration across its ecosystem at WWDC 2025, even as its enhanced Siri announced last year remains unfinished. New features include Live Translation for real-time language translation in phone calls, FaceTime, and iMessage with on-device processing to maintain privacy. Visual Intelligence can now analyze on-screen content through screenshots, enabling product searches on Google or ChatGPT queries about images.

The company introduced Workout Buddy, an AI-powered coach for Apple Watch that generates personalized voice coaching based on real-time biometric data.

Apple is opening its on-device AI models to developers for the first time, allowing offline functionality in apps. The company also unveiled a new glassmorphism design language across all operating systems with shiny, transparent app icons and menus.

🧠💰 OpenAI Launches o3-pro with 80% Price Cut

OpenAI released o3-pro, an upgraded reasoning model that outperforms competitors while slashing prices by 80%. The model excels at PhD-level math and science tasks, offering "4/4 reliability" testing and full tool access including web search and data analysis. Available immediately for Pro and Team users, it represents OpenAI's most capable reasoning model to date.

⏰📦 OpenAI Delays Open Model Launch to Summer

Sam Altman announced OpenAI's first open-weight model in years won't launch until late summer, pushed back from June due to a "very worth the wait" research breakthrough. The model represents OpenAI's attempt to re-engage the developer community and compete with open offerings from Mistral, DeepSeek, and Qwen.

⚖️💬 OpenAI Fights Court Order to Preserve All User Chats

OpenAI is challenging a court mandate from The New York Times lawsuit requiring retention of all user conversations, including deleted chats. The order affects hundreds of millions of users across all tiers, with CEO Sam Altman proposing "AI privilege" similar to doctor-patient confidentiality. Enterprise and API customers with Zero Data Retention agreements are exempt.

💵📈 OpenAI Hits $10B Annual Revenue Milestone

OpenAI reached $10B in annual recurring revenue, nearly doubling from last year. The company projects revenue of $125B by 2029, marking explosive growth in the AI sector.

🧸🤖 OpenAI Partners with Mattel for AI-Powered Toys

OpenAI and Mattel announced a partnership to create AI-powered toys across brands like Barbie, Hot Wheels, and American Girl. The first AI-powered product is expected later this year, with both companies emphasizing safety and age-appropriate design while granting Mattel employees access to ChatGPT Enterprise.

🌟🧠 OpenAI CEO: AI Takeoff Has Started

Sam Altman published "The Gentle Singularity," claiming humanity has passed the AI event horizon. He predicts AI creating new ideas by 2026, functional robots by 2027, and abundant intelligence and energy in the 2030s. Altman envisions a manageable transition where society adapts to exponential progress.

💭❤️ OpenAI Addresses Human-AI Relationships

OpenAI's Head of Model & Behavior Policy Joanne Jang detailed the company's approach to human-AI bonds and consciousness questions. The company focuses on warm, helpful personalities without fictional backstories while acknowledging people naturally anthropomorphize AI. This reflects broader questions about digital companionship's impact on society.

☁️🤝 OpenAI to Rent AI Chips from Google Cloud

OpenAI plans to rent AI chip servers from Google Cloud despite their rivalry in AI development. The move follows OpenAI's break from exclusive server deals with Microsoft as it desperately seeks to expand compute capacity. It's unclear whether OpenAI will rent Nvidia chips or Google's tensor processing units.

🔍📉 Google AI Slashes Traffic to News Publishers

Google's AI Overviews and chatbots are decimating referral traffic to media outlets, with major publishers like The New York Times seeing 8+ percentage point drops since 2022. As AI-powered summaries scrape content for instant answers, publishers lose ad revenue and turn to licensing deals or revenue-sharing models to survive.

🎙️🔊 Google Tests AI-Generated Podcast Summaries

Google is experimenting with "Audio Overviews" that transform search results into AI-generated podcast-style conversations. Available through Labs in the US, the feature creates brief dialogues between synthetic voices, offering hands-free information consumption with links to sources.

🌦️📊 Google DeepMind Launches Weather Lab

Google DeepMind launched Weather Lab, an interactive platform showcasing AI-powered weather forecasts for early, accurate predictions of storm paths and intensity. The tool demonstrates DeepMind's advances in meteorological AI applications.

🇫🇷☁️ Nvidia Partners with Mistral for European Cloud

Nvidia partnered with France's Mistral AI to launch sovereign cloud infrastructure in Europe, with major French companies already signed on. Nvidia also plans Germany's first industrial AI cloud platform, combining AI and robotics for automakers like BMW, boosting the European AI ecosystem.

🧮🚀 Mistral Debuts "Magistral" Reasoning Models

Mistral launched Magistral, a family of reasoning models for multi-step math, logic, and decision-making. Available in Small (24B, open source) and Medium (API) sizes, the models promise 10x faster performance in Le Chat with multilingual support, though they lag behind Claude Opus 4 and Gemini 2.5 Pro.

⚙️💻 Mistral Launches AI Infrastructure Stack

Mistral launched Mistral Compute, an AI stack offering GPU access, orchestration, and model training services. The move positions Mistral as an alternative to major cloud providers in the AI infrastructure space.

🧬🔬 Meta Creates 'Superintelligence' Lab with Scale AI

Meta is forming a new "superintelligence lab" led by Scale AI CEO Alexandr Wang through a $14.3B deal giving Meta 49% of Scale. Zuckerberg personally recruited nearly 50 researchers with nine-figure packages, following reported frustration with Llama 4's performance and desire to accelerate past competitors.

🌍⚙️ Meta's V-JEPA 2 Learns Real-World Physics

Meta released V-JEPA 2, a "world model" that understands physics and predicts real-world outcomes. Trained on 1M+ hours of video, it achieved 65-80% success rates manipulating unfamiliar objects, running 30x faster than Nvidia's Cosmos while humans score 85-95% on physical reasoning tasks.

⚖️🚫 Meta Sues AI "Nudify" App for Ad Abuse

Meta filed a lawsuit in Hong Kong against Crush AI, an app using generative AI to strip clothing from images. The company allegedly ran over 8,000 deceptive ads on Facebook and Instagram, evading detection with renamed domains and fake accounts.

✂️🎬 Meta Adds AI Video Editing to Meta AI App

Meta announced new AI video editing capabilities in its Meta AI app, allowing users to quickly change outfits, locations, lighting, and more with preset prompts. The feature expands Meta's consumer AI offerings with practical creative tools.

🎥⭐ Midjourney Launches Video Ranking Feature

Midjourney launched video ranking, allowing users to explore and rate outputs from its soon-to-be-released video model. The feature prepares the community for Midjourney's expansion into AI video generation.

🏰⚖️ Disney and Universal Sue Midjourney

Disney and Universal filed Hollywood's first major lawsuit against an AI startup, suing Midjourney for copyright infringement. The studios claim Midjourney scraped their iconic characters to train its models, allowing users to generate unauthorized knockoffs of characters like Yoda, Spider-Man, and Minions.

☁️🤖 AWS Revamps AI Services to Compete

Amazon Web Services is revamping Bedrock runtime to make it easier for businesses to develop AI apps and agents. The update aims to catch up with Google and Microsoft by supporting a wider array of AI models and simplifying development of agentic applications.

🌐🤖 The Browser Company Releases AI-First Dia Browser

The Browser Company launched Dia browser in beta, featuring an integrated AI chatbot that can see every tab and take agentic actions. The AI analyzes multiple tabs, drafts emails in user's style, and uses "Skills" for specialized tasks, with all data encrypted locally.

📊👥 Databricks Targets Non-Technical Users

Databricks unveiled a strategy shift with new products for business teams, including a drag-and-drop interface for building charts and data pipelines. The $62B company aims to expand beyond technical users to finance and marketing managers, competing with rival Snowflake.

📚🛑 Wikipedia Halts AI Summary Test

Wikipedia paused a pilot adding AI-generated summaries atop articles following editor criticism. The "unverified" summaries were only shown to opt-in users via browser extension, reflecting tensions between open-source communities and AI-generated content over factual accuracy concerns.

🏀🤖 Kalshi Airs AI-Generated NBA Finals Commercial

Prediction market Kalshi aired one of the first AI-generated commercials during NBA Finals Game 3, created with Google's Veo 3 in just 2 days. The 30-second spot cost 95% less than traditional production, marking a major moment for AI in advertising. Watch the video.

🎬🇨🇳 ByteDance's Seedance 1.0 Tops Video Leaderboards

ByteDance released Seedance 1.0, ranking first on video generation benchmarks ahead of Google's Veo 3, Kling 2.0, and Sora. The model generates 5-second, 1080p videos in under a minute with multi-shot storytelling and character consistency.

🧠🇨🇳 China's MiniMax to Release Open-Source Reasoning Model

MiniMax plans to unveil M1, an open-source AI reasoning model challenging DeepSeek's R1. The Shanghai-based company, valued at $3B with backing from Alibaba and Tencent, aims to intensify competition in the global open-source AI market.

🇨🇳💻 Huawei Founder: US Overestimates Our Chip Capabilities

Huawei founder Ren Zhengfei said the US "overestimates" his company's chip design capabilities, admitting Ascend AI chips lag US offerings by a generation. Despite growing domestic demand due to export controls, large tech companies hesitate over technical concerns with Huawei chips.

🎓🚫 Chinese AI Giants Block Tools During National Exams

Major Chinese tech companies temporarily blocked AI features during gaokao university entrance exams to prevent cheating among 13M+ students. ByteDance's Doubao, DeepSeek, Qwen and others suspended photo recognition and exam-related queries during June 7-10 testing periods.

🎓🤖 Ohio State Launches AI Fluency Initiative

Ohio State University is launching an AI Fluency Initiative to embed AI education in undergraduate programs. The initiative includes resources, courses, and support for faculty and students to integrate AI across academic disciplines.

📜🔬 AI Reveals Dead Sea Scrolls Are Century Older

Researchers discovered Dead Sea Scrolls may be up to a century older using an AI system called Enoch that analyzes handwriting patterns. The non-destructive method pushes some texts back 2,300 years, offering new insights into biblical manuscripts without damaging precious documents.

🏗️🇬🇧 UK Uses Gemini to Fast-Track Infrastructure Planning

UK government and Google unveiled "Extract," an AI tool using Gemini to digitize millions of planning documents. The tool converts 2-hour manual processes into 40 seconds, with nationwide rollout by Spring 2026 to help meet 1.5M home-building targets.

⚖️⚠️ UK Court Warns Lawyers About Fake AI Citations

The High Court of England and Wales warned lawyers submitting AI-generated fake case citations risk severe sanctions. Judge Victoria Sharp emphasized ChatGPT cannot be relied on for legal research without human verification, with penalties ranging from fines to police referral.

Community

Join AI Whisperer Community!

Ready to take your AI journey to the next level? Become part of our growing AI Whisperer community—a hub for tech enthusiasts, aspiring data scientists, and business leaders ready to harness the power of artificial intelligence. Inside, you’ll find:

• Community: Like-minded individuals keen to grow together

• Curated AI Tools: Discover the latest and greatest tools to boost productivity.

• Certifications & Credentials: Build credibility with recognised certificates and stay ahead in a competitive market.

Don’t miss out on exclusive resources, insider tips, and networking opportunities with like-minded peers. Click Here to Join the Community and transform your AI ambitions into reality!

That's it for this week!
Until next time, stay curious and keep exploring the ever-evolving world of AI!
Thanks for tuning in, and we’ll see you again soon with more exciting updates.

Jul