📜 Microsoft in Talks to Lock In Post-AGI Access to OpenAI
Microsoft is negotiating a new deal with OpenAI to retain access to its models, even if OpenAI declares it has reached artificial general intelligence (AGI). Under the current contract, AGI would limit Microsoft’s rights. The revised agreement, still under discussion, reflects growing tensions as OpenAI also courts other cloud providers like Google and Oracle.
🎬 NotebookLM Adds Video Overviews and Smarter Studio
NotebookLM now features Video Overviews, AI-generated slide presentations that visually explain complex topics using your documents. The revamped Studio panel lets users create multiple outputs like Audio Overviews, Mind Maps, and Reports per notebook. It's designed for deeper learning, collaboration, and customization.
👩🏫 ChatGPT Launches Interactive Study Mode for Students
OpenAI has launched "study mode" in ChatGPT to promote deeper learning through Socratic questioning and step-by-step guidance. Rather than just giving answers, it helps students work through problems interactively. Available now to Free, Plus, Pro, and Team users.
🏛️ China Proposes Global AI Governance as U.S. Pursues Solo Strategy
At WAIC 2025, Chinese Premier Li Qiang proposed a global AI governance framework, urging collaboration over monopolization. The move contrasts sharply with the Trump administration’s new plan to secure U.S. dominance in AI. As tensions rise, China’s tech sector is quickly gaining ground with cost-effective, high-performing AI models.
Enjoying our newsletter? Forward it to a colleague—
it’s one of the best ways to support us.
👾 FORWARD FUTURE ORIGINAL
Perplexity CEO on Building Comet Browser, Beating Google, and the Future of AI Agents
He's building the company that could dethrone Google. Aravind Srinivas, CEO of Perplexity AI, joins for an in-depth conversation about the company's audacious browser launch, their strategy to compete with tech giants, and why the future of web browsing will be powered by AI agents.
In this revealing interview, Aravind explains why Perplexity built their own browser from scratch in just 8 months, how they're avoiding the advertising trap that could destroy Google's business model, and his candid thoughts on AI's impact on jobs. → Watch the full interview here.
🕵️♂️ ESPIONAGE
US and China Race to Deploy Cutting-Edge AI in Intelligence Ops
The U.S. intelligence community is accelerating efforts to adopt advanced AI after being blindsided by China's rapid deployment of large language models like DeepSeek. In a $200M push, the Pentagon has contracted OpenAI, Anthropic, Google, and xAI to experiment with "agentic" models—systems that can autonomously break down tasks and control other tools. These models are already being fine-tuned for top-secret use, with agencies deploying secure versions tailored for language, reliability, and classified environments.
But despite the tech, progress is slow. Intelligence officials worry the U.S. is bogged down by bureaucracy, while China, with fewer constraints, may leap ahead. Experts warn that without a culture shift in how national security agencies operate, America risks winning the AGI race but losing the one that actually matters: adoption. → Read the full article here.
🎓 EDUCATION
12,000 Prompts, No Judgment: What Student ChatGPT Logs Reveal
After 18 months and nearly 12,000 prompts, a journalist examined the ChatGPT usage of three British undergrads, and uncovered a candid, sometimes troubling portrait of modern student life. The logs show students using the chatbot not just for academic support, but for career advice, mental health guidance, and even emotional companionship. It’s a kind of digital outsourcing of decision-making, where AI becomes tutor, therapist, and best friend rolled into one.
The students’ dependence is both practical and existential. With academic pressure high and human support uneven, ChatGPT offers answers without judgment—though not always with accuracy. The logs expose how AI shapes self-perception and behavior, offering comforting affirmations and plausible-sounding but occasionally flawed advice. → Read the full article here.
🧫 SCIENCE
AI Maps Gut Bacteria's Chemical Chatter, Paving New Paths for Personalized Medicine
Researchers at the University of Tokyo have developed an AI tool, VBayesMM, that uses a Bayesian neural network to untangle the chemical signals produced by the trillions of bacteria in our gut. By analyzing microbiome and metabolite data, the system reveals which microbes drive key health-related chemicals—and does so with a refreshing dose of statistical humility.
Tested across studies on sleep disorders, obesity, and cancer, VBayesMM outperformed conventional methods, pinpointing bacterial families tied to known biological roles. While challenges remain, like computational cost and the tangled interactions between microbes, the goal is clear: harness gut chemistry for precision treatments. → Read the full paper here.
🛰️ NEWS
What Else is Happening
🦿 Skild AI Unveils Robot Brain: Backed by Amazon, Skild’s new AI model lets robots walk, sort, and adapt like humans, tackling robotics’ data gap with shared learning across machines.
🧐 Google Search Gets Smarter: New AI Mode tools like Canvas and Search Live bring study guides, visual Q&A, and PDF uploads, turning search into an interactive tutor with real-time context.
📈 Anthropic Eyes $170B Valuation: Iconiq Capital is leading a $5B funding round for AI rival Anthropic, nearly tripling its value, and fueling the race for AI dominance beyond Silicon Valley.
🚀 Enfabrica Tackles AI Memory Costs: Backed by NVIDIA, Enfabrica’s new EMFASYS system links cheaper DDR5 memory to AI chips—aiming to cut pricey HBM reliance without slowing performance.
👔 Meta Poaches Fourth Apple AI Star: Apple loses key researcher Bowen Zhang to Meta’s superintelligence team, deepening turmoil in its AI unit as rivals offer massive pay to lure top talent.
🎧 Spotify Teases Chatty AI Future: Voice requests are just the start, Spotify plans a more conversational, reasoning AI that deeply understands your taste, powered by playlist and voice data insights.
🧰 TOOLBOX
Superhero Merch, Smarter Resumes, and Fast-Tracking Visual AI
Shop licensed superhero gear, exclusive designs, and collectibles with global shipping. | Build standout resumes, cover letters, and sites fast with AI, templates, and grammar help. | Prototype and scale visual AI fast with streamlined MLOps and data for top accuracy. |
🤠 THE DAILY BYTE
Meet TARS, the Sun-Powered Light Sail Launching Chips to the Stars
New paper! Allow me to introduce TARS = Torqued Accelerator using Radiation from the Sun. TARS (yes inspired by Interstellar!) is a rotating light sail that's capable of launching chip sats into interstellar space using only radiation from the Sun, so let's dive into how it works — P(David|Kipping) ∝ P(Kipping|David) P(David) (@david_kipping) 4:17 PM • Jul 26, 2025
@david_kippingView post on X
