The Web Is Drawing a Line Between AI Training and Everything Else
Today’s AI news circles a single question that’s been building for years: who actually controls the internet’s content, and who gets to profit from it? Cloudflare just forced that question into the open with a hard deadline. Meanwhile, investors are making big bets on the infrastructure beneath all these AI systems — and Google is quietly expanding what an AI assistant can do while you work.
Cloudflare Tells AI Companies: Separate Your Crawlers or Get Blocked
Most people have never thought about web crawlers — the automated programs that AI companies use to scan and collect text from websites across the internet. Cloudflare, which manages internet security and traffic for roughly 20% of all websites, just put those crawlers at the center of a fight over money and control.
Starting September 15, Cloudflare will require AI companies to use separate, clearly identified crawlers for AI training versus standard web indexing. Right now, many companies bundle both functions into a single tool, which makes it nearly impossible for publishers to say yes to being indexed by search engines while saying no to having their articles fed into an AI model’s training data. Cloudflare’s new policy forces them to split those two activities apart.
Think of it like a restaurant that lets food critics write reviews but doesn’t want a competitor photographing every dish on the menu. The problem was that both visitors looked identical at the door. Cloudflare is now requiring name badges.
For everyday readers, this matters because the websites you read every day — news outlets, recipe blogs, forums — have been struggling to survive in an era when AI systems summarize their content without sending traffic back to them. This policy gives publishers a real lever to pull. If an AI company doesn’t comply by September 15, Cloudflare can block their crawlers entirely. TechCrunch has the full breakdown.
Why this matters: This is one of the most concrete moves yet to give content creators bargaining power over AI companies. A hard deadline with real enforcement changes the negotiation.
“Cloudflare deadline of September 15 for separating AI training crawlers from search crawlers”
Ashton Kutcher Is Betting Big on AI’s Electricity Problem
Venture capital — the business of investing early in startups in exchange for ownership stakes — is reorganizing itself around AI. The latest sign: Ashton Kutcher is leaving Sound Ventures, the firm he co-founded eleven years ago, to launch a new fund with Morgan Beller focused specifically on AI infrastructure and power systems.
Beller is not a newcomer. She co-created Facebook’s Libra cryptocurrency project and later joined Andreessen Horowitz. The two are targeting the unglamorous but essential layer of AI: the data centers, energy systems, and hardware that make large AI models actually run. Chips and algorithms get the headlines, but without reliable, affordable power, none of it works.
The analogy here is the California Gold Rush. Most fortunes weren’t made mining gold — they were made selling shovels and blue jeans. Kutcher and Beller seem to be betting that the real money in AI’s next phase is in the picks and shovels: power grids, cooling systems, and the physical infrastructure underneath the models everyone talks about.
For ordinary people, this signals where serious money thinks AI is heading. The flashy chatbot era may be maturing, and the next wave of competition will happen in power and infrastructure. That affects energy prices, local economies near data centers, and ultimately the cost of every AI product you use. The Wall Street Journal confirmed the move, with additional reporting from TechCrunch.
Why this matters: When experienced investors exit established funds to start something new, it signals a real shift in where the opportunity is. AI infrastructure is quietly becoming the most contested space in tech.
Google’s Gemini Spark Comes to Mac — and It Watches Your Screen
Agentic AI — software that takes actions on your behalf rather than just answering questions — just became a little more present in everyday computing. Google released Gemini Spark for Mac as a beta feature, available now to Google AI Ultra subscribers in the United States.
Spark is designed to run continuously in the background, tracking what you’re working on in real time and connecting across apps to help you without being explicitly prompted. It’s less like asking a question and more like having an assistant who’s already watching what’s on your desk. The Mac release includes expanded app compatibility, meaning it can interact with more of the software people actually use day to day.
This puts Google directly in competition with Apple’s own on-device AI features and Microsoft’s Copilot, which has been embedded in Windows for over a year. The beta is limited for now, but the direction is clear: these companies want AI running constantly on your computer, not just when you open a chat window. Engadget covered the release alongside TechCrunch.
Why this matters: The shift from AI as a tool you open to AI as something always running is a significant change in how these products work — and how much access they have to your daily life.
Also Happening in AI
Privacy-focused AI startup Venice AI hit unicorn status — a valuation of $1 billion or more — after raising $65 million in a Series A round, a sign that there’s real appetite for AI platforms that don’t store your conversations, according to TechCrunch. Meta is moving in a different direction entirely: the company is reportedly building a cloud business to sell its excess AI computing capacity to other companies, following a path similar to what SpaceX has done with Starlink infrastructure, per Reuters. On the enterprise side, device management company Jamf launched an AI governance tool for Mac, helping businesses track and control which AI apps employees are running — a quiet but telling sign of how seriously companies are now treating AI oversight, as IT Brief UK reports. Anthropic also pushed out a routine maintenance update to its Python SDK, version 0.115.1, tidying up some unused code under the hood.
What to Watch
The Cloudflare deadline creates a forcing function that the whole industry will have to respond to publicly by mid-September — watch for which AI companies comply quickly, which ones push back, and whether any major publishers start announcing licensing deals as a result. The parallel stories of Meta building cloud infrastructure and Kutcher’s new fund targeting AI power systems suggest that the next big AI competition won’t be about which chatbot is smarter. It’ll be about who controls the energy and compute underneath them.