Weekly AI News

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

And what a week it’s been!
President Donald Trump wasted no time repealing Joe Biden’s AI executive order, igniting debates on whether speed or safety should take the lead in AI development. Meanwhile, OpenAI stole the spotlight with a jaw-dropping $500 billion investment plan to supercharge U.S. AI infrastructure—a bold step to secure America’s tech dominance. Perplexity AI shook things up with an audacious bid to merge with TikTok U.S., blurring the lines between AI-powered search and social media.

But that’s not all—Anthropic teased smarter models and breakthrough features, while Google DeepMind’s Gemini 2.0 “Flash Thinking” is already setting new benchmarks in reasoning and science. Across the globe, ByteDance and DeepSeek continue their rapid ascent, proving China is far from lagging in the AI race. And let’s not forget: humanoid robots assembling iPhones and the UK’s cutting-edge supercomputer redefining vaccine research are driving the future of innovation in ways we never imagined.

The pace of progress is dizzying, and we’re just getting started. Buckle up—AI is not just transforming the world; it’s rewriting the rules of what’s possible.

🚀🇺🇸 OpenAI Announces $500B U.S. AI Infrastructure Initiative

OpenAI unveils “The Stargate Project,” partnering with SoftBank, Oracle, and other giants to build massive data centers across the country. The initial $100B investment targets Texas first, aiming to create hundreds of thousands of jobs. Trump’s repeal of Biden’s AI order sparks debate on prioritizing speed over safety.

🤖🌐 OpenAI Launches Its First Autonomous Web Agent

Called “Operator,” the AI system navigates websites independently to handle tasks like booking reservations and ordering groceries. It combines vision with advanced reasoning and partners with platforms such as DoorDash and Uber. Safety features include purchase approvals, threat detection, and user takeover for sensitive data.

🏎️💡 OpenAI Readies ‘o3-mini’ Model Release

CEO Sam Altman confirms a near-future launch for the o3-mini reasoning model, available via both API and ChatGPT. Positioned as faster but less capable than o1 Pro, o3-mini highlights OpenAI’s push for rapid AI model expansions. Altman hints that “o3” and “o3-pro” could soon arrive in the $200/month ChatGPT Pro tier.

🎓⚡ Altman to Brief Washington on ‘PhD Level SuperAgents’

Sam Altman will present advanced AI that can solve complex problems at an expert level to U.S. officials on Jan. 30. Internal reports indicate GPT-4b micro can engineer proteins 50x more effectively than scientists. These behind-the-scenes models highlight how OpenAI’s biggest breakthroughs are happening off the public radar.

🤝📈 OpenAI CPO Kevin Weil Hints at Next-Gen Model Training

Speaking at Davos, Weil revealed that OpenAI is already training the successor to the forthcoming o3 reasoning model. Despite compressed release cycles, he anticipates “another big jump in capabilities.” The rapid cadence underscores OpenAI’s commitment to frequent, impactful AI updates.

☁️🔒 Microsoft Amends Exclusive Cloud Agreement With OpenAI

Microsoft retains first-refusal rights on OpenAI’s computing capacity but allows the AI powerhouse to seek other infrastructure partners. This adjustment expands OpenAI’s deployment options while keeping Microsoft in a privileged position. It reflects the evolving nature of high-stakes cloud collaborations.

💡🤳 Perplexity AI Makes a Bid to Merge With TikTok U.S.

Perplexity AI proposes a $50+ billion merger to combine advanced AI search with TikTok’s massive user base. This move could introduce AI-driven experiences while addressing U.S. ownership and national security concerns. Skeptics question data privacy and potential regulatory hurdles.

📱🗣️ Perplexity Debuts New AI Mobile Assistant

A free agent-like tool for Android, Perplexity Assistant can control phone apps (e.g., Uber, OpenTable) via voice or gestures. It supports multimodal inputs and remembers context for seamless transitions from research to action. Free on Android now, it directly challenges mainstream voice assistants.

 🔎🌐 Perplexity Launches Sonar API for Real-Time Search

Sonar and Sonar Pro APIs offer citation-backed search capabilities for developers, delivering credible and current information. Sonar Pro handles complex queries with more citations and a larger context window. These new APIs aim to reduce hallucinations and build user trust in AI-driven search.

⚡🌌 Google DeepMind Debuts Gemini 2.0 ‘Flash Thinking’

A free experimental AI surpassing math and science benchmarks, it claims top spot on LM Arena’s leaderboard. A 1M-token context window plus built-in code execution elevate reasoning power. Beta users enjoy it at no cost, while OpenAI charges $200/month for premium reasoning access.

 💸🙋 Google Reportedly Invests More Than $1B in Anthropic

According to the Financial Times, Google is injecting an additional $1B into Anthropic, after already committing $2B in 2023. Anthropic has also raised $8B from Amazon and is seeking $2B more led by Lightspeed. This signals escalating investor faith in Anthropic’s AI research and cloud partnerships.

🗣️🧠 Anthropic CEO Dario Amodei Talks Future of Voice and Memory

At Davos, Amodei revealed a two-way voice mode for Claude, plus features to let Claude remember user context. He expects “smarter” AI models in the coming months, fueled by over 1M chips by 2026. Anthropic faces “overwhelming” demand as it scales to meet enterprise needs.

📚🗒️ Anthropic Launches Citations API for Enhanced AI Transparency

Claude can now reference specific sentences and passages in source documents, reducing hallucinations and boosting accountability. Available via the Anthropic API and Google Cloud’s Vertex AI, it targets use cases like document summarization and Q&A. Clients like Thomson Reuters applaud the improved accuracy.

 🇫🇷🚀 Mistral CEO Arthur Mensch Denies Acquisition Rumors

Already valued near $6B, the French AI startup is exploring an IPO instead of selling. Mensch says Mistral remains committed to independence and continued innovation. Despite industry speculation, Mistral’s focus is on scaling its own cutting-edge AI rather than a merger or buyout.

🆓🐋 DeepSeek’s Open-Source R1 Beats OpenAI’s o1

Chinese lab DeepSeek releases a 671B-parameter reasoning model that outperforms o1 on math, physics, and coding tasks. R1’s MIT license allows commercial use, and costs are far lower than o1’s API pricing. Reinforcement learning proved key, letting the AI learn to “think step-by-step” without massive curated data.

🐼🤔 ByteDance Releases New Model and Reasoning Agent

Doubao 1.5 Pro excels in knowledge, coding, and reasoning, rivaling GPT-4o and Claude 3.5 Sonnet. ByteDance also open-sources veRL, the RL library behind its advanced reasoning. Another launch, UI-TARS, is an AI agent that interprets screenshots for dynamic UI interactions.

🏯🌐 Tencent Unveils Advanced 3D Generation AI

Hunyuan3D 2.0 open-sources a pipeline to generate and animate high-quality 3D assets. It splits shape creation (Hunyuan3D-DiT) and texturing (Hunyuan3D-Paint) for detailed, lifelike results. Hunyuan3D-Studio supports features like sketch-to-3D and stylization, streamlining game and film workflows.

🎨🔄 Runway Releases ‘Frames’ Image Generation Model

An addition to Runway’s creative suite, Frames produces photorealistic or stylistic images while preserving a consistent aesthetic (“Worlds”). Edits include seed settings, aspect ratio changes, and style variations. Paid users can integrate images directly into Runway’s video tools for cohesive content creation.

✂️💥 Adobe’s New AI Tool Edits 10,000 Images in One Click

Firefly Bulk Create can remove, replace, or extend backgrounds for large image batches. This streamlines workflows for marketers, photographers, and designers. Additional AI features from Adobe include automated video translation and resizing, marking a new era of high-volume content editing.

🎞️🚀 Netflix Unveils AI Video Generator ‘Go-with-the-Flow’

Features like Cut-and-Drag Animation and First Frame Editing enable smooth, consistent motion and easy object manipulation. Turntable Motion Transfer and Davis Motion Transfer replicate 3D camera perspectives or combine existing motion with new prompts. From relighting to super-resolution, Netflix is raising the bar in AI-driven animation.

⏩🎨 Midjourney Announces February V7 Release

Midjourney is preparing new partnerships for two video models, plus 3D capabilities and faster processing. The upcoming V7 builds on Midjourney’s popular image synthesis, offering creators expanded tools. It signals a push to remain competitive in the rapidly evolving AI art and video space.

🧠🐦 Hugging Face Unveils SmolVLM 256M and 500M

Dubbed the smallest vision-language models with competitive performance, SmolVLM reduces hardware demands. This step toward lightweight AI broadens potential applications in mobile and embedded devices. Hugging Face aims to democratize AI further by lowering resource barriers.

💻⚒️ Cognition Labs Releases Devin AI 1.2

The coding assistant gains improved context understanding, browser-based workplace setup, enterprise accounts, and Slack audio integration. These updates streamline collaboration for development teams. Devin AI continues to push advanced features for smoother, more efficient programming experiences.

📱🌟 Samsung Unveils Galaxy S25 Phones with AI Upgrades

At its “Unpacked” event, Samsung introduced enhanced Gemini integrations, multimodal agent features, and context-aware interactions. The S25 series aims to offer richer, more intuitive AI experiences. The company’s focus on user-centric design highlights the growing importance of integrated AI in smartphones.

🎮👾 AI Startup Character AI Tests Games on the Web

Character AI, known for chatbot-based characters, is adding mini-games to boost user engagement. Two games, Speakeasy and War of Words, let players challenge AI characters under certain constraints. Available to paid subscribers and select free users, it merges entertainment with conversational AI.

💼🤖 Goldman Sachs Introduces GS AI Assistant

Rolled out to 10,000 employees, this internal tool helps summarize emails, proofread texts, and translate code. Goldman aims to extend it to all knowledge workers by year’s end. The bank sees the assistant as a “virtual employee,” centralizing expertise and boosting productivity.

⚖️💬 LinkedIn Faces New Class-Action Lawsuit

The complaint alleges LinkedIn used premium subscribers’ private messages to train AI models without proper consent. This raises fresh concerns over data privacy and corporate accountability. The outcome could set precedent for how user communications are leveraged in AI training.

🤖📱 Humanoid Robots Assemble iPhones in China

Foxconn partners with UBTech to deploy the Walker S1 robot for iPhone assembly, performing tasks from quality checks to component fittings. Standing 5’6” and weighing 168 pounds, Walker S1 aims to reduce repetitive strain on human workers. UBTech strives for commercial-scale humanoid production.

🔄🏛️ Trump Repeals Biden’s Executive Order on AI

President Donald Trump scrapped the regulation requiring AI labs to share model details with the federal government. Critics say it accelerates U.S. AI development at the cost of oversight and safety. Supporters claim it removes bureaucratic red tape stifling innovation.

🏆📜 ‘Humanity’s Last Exam’ Scales Up AI Benchmark

The Center for AI Safety and Scale AI launched a 3,000-question test spanning 100+ subjects to challenge top-tier models. Leading AIs barely reach 10% accuracy, despite acing previous benchmarks. A $500k prize pool encourages question submissions to keep pushing AI evaluation forward.

🇮🇱💻 Google Provided AI Tools to Israeli Military

As part of Project Nimbus, Google granted Israel’s Ministry of Defense and the IDF access to Vertex and Gemini AI tech. Internal discussions reveal employees pressured for quick approval to avoid losing ground to Amazon’s services. The move sparked controversy after protests and employee firings at Google.

💊💉 UK’s Supercomputer Develops AI Vaccines

Isambard-AI, a $276M supercomputer in Bristol, ranks among the world’s top 10 fastest and focuses on drug and vaccine discovery. It tests millions of molecular interactions virtually to identify promising treatments for Alzheimer’s and heart disease. Excess heat from the system will help warm local buildings, showcasing eco-conscious planning.

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