Search the archive

now

AI REFERRALS TO TOP WEBSITES WERE UP 357% YEAR-OVER-YEAR IN JUNE, REACHING 1.13B

AI referrals to websites surged 357% year-over-year in June, reaching 1.13 billion, with ChatGPT responsible for over 80% of these referrals.

BUILD YOUR OWN AI APP BUILDER WITH THE V0 PLATFORM API

The v0 Platform API provides developers with a text-to-app interface, enabling automation of web app creation and code generation using LLM-generated UIs.

XAI TESTS TEAM ACCOUNTS FOR GROK AS COMPETITION HEATS UP

xAI appears to be testing a Teams plan for Grok that will let users toggle between personal and dedicated Team accounts.

RELEASE: POWER ATTENTION

Manifest AI has open-sourced Power Attention, which replaces Flash Attention in any transformer.

Meet our AI teams and discover your next role >>

CHINA CLAIMS NVIDIA BUILT BACKDOOR INTO H20 CHIP DESIGNED FOR CHINESE MARKET

China's cyber regulator recently held a meeting with Nvidia over serious security issues with the company's artificial intelligence chips. It claims that Nvidia's H20 chips, which were designed for the Chinese market to comply with US export restrictions, have location tracking and that they can be remotely shut down. Major Chinese tech groups have received informal guidance to increase purchases of domestic AI chips to reduce reliance on Nvidia and support the evolution of a rival domestic chip ecosystem.

PUBLIC CHATGPT CONVERSATIONS ARE APPEARING IN GOOGLE SEARCH

Shared ChatGPT conversations are being indexed by search engines like Google and Bing, revealing private or sensitive queries that users may not expect to be discoverable.

OPENAI LAUNCHES STARGATE NORWAY TO POWER SUSTAINABLE AI IN EUROPE

OpenAI launched Stargate Norway, its first AI data center in Europe, as part of the OpenAI for Countries program. This facility will deliver 230MW capacity and 100,000 NVIDIA GPUs by 2026, leveraging Narvik's hydropower and cool climate for sustainable AI infrastructure.

2025 MID-YEAR LLM MARKET UPDATE: FOUNDATION MODEL LANDSCAPE + ECONOMICS

Foundation models are shaping the future of computing. Model API spending has more than doubled in six months. Enterprises are increasing production inference rather than just model development, marking a shift from previous years. Code generation has become AI's first breakout use case. While open source continues to advance, Western labs have seen a slowdown in frontier breakthroughs, resulting in enterprise dollars consolidating around a few high-performing, closed-source models.

BUILDING AN AI SHOPPING ASSISTANT

A walkthrough for building an AI assistant that can browse online clothing stores, locate specific garments, and generate virtual try-on previews showing how the clothes would look on you.

THE XENO SUTRA: CAN MEANING AND VALUE BE ASCRIBED TO AN AI-GENERATED “SACRED” TEXT?

Buddhist scholars analyzed an AI-generated "sutra" created by ChatGPT o3, finding sophisticated philosophical concepts and imagery that effectively evokes core Buddhist teachings like emptiness through paradoxical deconstruction. However, they warned that while LLMs can recite dharma teachings, they lack true insight and may dangerously convince users they're receiving true divine guidance when prompted for spiritual advice.

what's really happening behind the hype. Get the full breakdown

and learn how to improve DevEx in an AI-powered world.

LEARNING THE BITTER LESSON

The design philosophy for building AI applications is still in its infancy. One thing we can predict is that models will get much better. Designing AI applications to take advantage of this is probably the most important thing. It's important to understand your application structure, re-evaluate that structure as models improve, and make it easy to remove structure.

NEURAL NETWORK COMPRESSION

A Fractional Gaussian Filter and Pruning (FGFP) framework combines fractional-order differential calculus and Gaussian functions to compress neural networks for edge deployment.

FACING QUESTIONS ON AI STRATEGY, TIM COOK SAYS APPLE IS ‘VERY OPEN' TO ACQUISITIONS

Apple CEO Tim Cook's remarks on the company's third-quarter earnings call confirm that the company is going to invest more heavily in AI technology. He said that Apple is always looking to buy companies of any size that could help it develop its AI offerings. Apple is facing growing pressure from Wall Street to catch up to its Silicon Valley peers. In addition to growing its investment, Apple is also rearranging staff internally to focus more on AI.

META SHARES JUMP AS AI FUELS AD SALES, OUTWEIGHING BIG CAPITAL COSTS

Meta's shares are up 11% after earnings today, driven largely by the success of AI-powered ad recommendations, which drove 5% more conversions on Instagram and 3% on Facebook. Mark Zuckerberg told analysts the company will push aggressively on AI investments as the most aggressive AI timelines have held true.

HOW AN ‘ENTREPRENEUR OF THE YEAR' BROUGHT THE FIRST BIG BUST TO AI BOOM

Sachin Dev Duggal was ousted as CEO of $1.5 billion AI unicorn Builder.ai after investigators uncovered $249 million in fabricated sales and "round tripping" schemes. The company, which claimed AI could automate app development but relied heavily on human programmers in India, allegedly overstated revenue by 300% while Duggal transferred investor funds to personal accounts for luxury expenses.

Subscribe now.

REVERSE ENGINEERING SOME UPDATES TO CLAUDE

Anthropic has released two major new features for Claude: the ability to draft and send emails, messages, and calendar invites directly from the Claude app, and the ability to upload PDFs, images, code files, and more to AI-powered apps.

GITHUB COPILOT REACHES 20 MILLION USERS

GitHub's AI coding assistant hit 20 million cumulative users, adding 5 million since April and now serving 90% of the Fortune 100, with enterprise adoption up 75% quarter‑over‑quarter.

PHOTO REMIX TOOLS IN GOOGLE PHOTOS

Google Photos has introduced new creative tools that let users transform still photos into short videos and remix images with artistic effects.

this topic: _The_ _Monetization Operation Model_. Get early access

OPENAI HITS $12 BILLION IN ANNUALIZED REVENUE

OpenAI has reached $12 billion in annualized revenue, roughly doubling its revenue in the first seven months of the year. The figure implies that OpenAI is generating $1 billion a month. The company has around 700 million weekly active users. OpenAI has increased its cash burn projection to roughly $8 billion in 2025, up $1 billion from its projection earlier in the year.

META'S VISION FOR PERSONAL SUPERINTELLIGENCE

Meta is betting on personal superintelligence as the future of computing: AI that knows you deeply, lives on devices like smart glasses, and helps you pursue your goals, relationships, and creativity. Unlike centralized visions of automation and universal basic output, Meta argues that progress comes from individual agency and empowerment. The company plans to invest heavily in infrastructure and safety to bring this technology to billions.

AMAZON INVESTS IN ‘NETFLIX OF AI' START-UP FABLE, WHICH LAUNCHES SHOWRUNNER: A TOOL FOR USER-DIRECTED TV SHOWS

Amazon has invested in Fable, a startup with an AI-generated TV show service called Showrunner. The money from the investment will help develop Showrunner, which allows users to create scenes (or entire episodes) of a TV show using prompts. Showrunner will be free at launch, but the company eventually plans to charge $10-$20 per month for credits. Fable's AI models have guardrails that block offensive, illegal, or copyrighted material.

MISTRAL CODESTRAL 25.08 AND THE ENTERPRISE CODING STACK

Mistral unveiled Codestral 25.08, part of a full-stack coding platform optimized for enterprise needs. It addresses deployment, customization, observability, and toolchain integration gaps to enable AI-native software development across regulated and complex environments.

QUERY FAN-OUT TECHNIQUE IN AI MODE: NEW DETAILS FROM GOOGLE

Google's query fan-out technique, available in AI Mode, Deep Search, and some AI Overview experiences, makes multiple background searches based on an initial question. It uses a large language model to interpret queries and 'fan out' to multiple related searches, including on topics the user never explicitly mentioned. The responses are then combined into a single response with links.

30% OF "HUMANITY'S LAST EXAM" ANSWERS ARE LIKELY WRONG

FutureHouse researchers found that nearly a third of biology and chemistry questions in the prominent PhD-level AI benchmark contain answers directly contradicted by peer-reviewed literature. The errors stem from HLE's unusual design that created "gotcha" puzzles that even confused human experts.

PRICING AI PROOFS-OF-CONCEPT: FREE PILOTS WILL KILL YOU

AI agents deliver real business value, so they should be priced accordingly from day one. Companies that position their products as pilots or technical proof-of-concepts are competing on features instead of outcomes. They should charge enough to filter out the curious and co-create the ROI model that closes the deal. Customers who see how agents help save time, reduce errors, and increase throughput won't be able to argue with the business value.

👉 See how top teams do it

AGENT INTERACTION GUIDELINES (AIG)

Agents are reshaping roles and workflows, changing how software is planned, built, reviewed, and deployed. Developer's roles have shifted to orchestrating input and reviewing output. The Agent Interaction Guidelines are a set of foundational, evolving principles and practices for designing agent interactions that integrate naturally into human workflows. They are still being written, and more is expected to be added as the community gains more experience.

GEPA: REFLECTIVE PROMPT EVOLUTION CAN OUTPERFORM REINFORCEMENT LEARNING

Traditional RL approaches need thousands of expensive trial-and-error runs to learn from simple success/failure scores. GEPA uses natural language reflection to diagnose what went wrong and evolve better prompts through a Pareto-based selection strategy. GEPA (Genetic-Pareto) dramatically outperforms both reinforcement learning methods like GRPO and leading prompt optimizer MIPROv2 while requiring up to 35× fewer training attempts.

LATTE: DECENTRALIZED TEST-TIME ADAPTATION

Latte is a framework for test-time adaptation in decentralized environments using pre-trained vision-language models. It combines local and shared memory to improve adaptation under limited data conditions.

SOMETHING THAT HAS BECOME UNDENIABLE THIS MONTH

The best available open weight models new come from Chinese AI labs. The delay in the release of OpenAI's open weights model may be due to a desire to be notably better than the truly impressive lineup of Chinese models. This post lists the biggest Chinese model releases of this month, along with links to notes on each one.

‘VIRTUAL SCIENTISTS' TO SOLVE COMPLEX BIOLOGICAL PROBLEMS

Stanford Medicine developed a team of AI agents that mimic real research labs, complete with a virtual PI, specialized scientists, and a built-in critic. The system has already designed promising COVID-19 nanobodies in days that outperformed existing antibodies in lab tests.

GOOGLE EARTH AI

Google introduced Google Earth AI, a suite of geospatial AI models and datasets designed to tackle major planetary challenges. It includes models for weather prediction, flood forecasting, wildfire detection, and urban planning.

MICROSOFT IN ADVANCED TALKS FOR CONTINUED ACCESS TO OPENAI TECH

Microsoft and OpenAI are negotiating new terms that would let Microsoft continue using OpenAI's technology even after the startup declares AGI— a crucial pivot since their current contract would cut Microsoft off from advanced models at that milestone.

ANTHROPIC BACKS HEALTHCARE INTEROPERABILITY

Anthropic joined a CMS initiative to modernize healthcare data sharing using AI.

META'S $72B BET ON AI INFRASTRUCTURE

Meta plans to spend up to $72 billion on AI infrastructure in 2025, doubling its capex to build data centers and superclusters like Prometheus and Hyperion.

👉 Take the assessment

ANTHROPIC IN TALKS TO RAISE FRESH CAPITAL AT $170 BILLION VALUATION

The funding round, led by Iconiq Capital, would nearly triple Anthropic's valuation from $61.5 billion just four months ago.

CHATGPT STUDY MODE

OpenAI launched Study Mode in ChatGPT to promote active learning by guiding students through problems with step-by-step prompts tailored to their goals and proficiency. This feature aims to support deeper understanding rather than just providing answers, and is now available to all logged-in users.