In partnership with
Domain Names + Web and Email Hosting Done Right
Tired of overpaying GoDaddy or Namecheap? There's a smarter option with Porkbun!
Porkbun is the domain registrar trusted by creators, developers, entrepreneurs, and anyone who wants low prices without the nonsense or the forced upsells.
Here's why so many people are making the move to the Bun:
• Most domains sold at cost
• Low, transparent registration and renewal pricing
• Free features like WHOIS privacy and SSL certificates
• Powerful web and email hosting options
• Real human support 24/7, 365 days a year
• Named the #1 domain registrar by Forbes Advisor and USA Today
Whether you're launching a business, building a personal brand, starting a side project, or creating your very first website, Porkbun makes the whole thing easy.
Anthropic just changed what happens to every piece of writing Claude produces, and it did not ask first. Nvidia and six of Wall Street's biggest names just bet half a trillion dollars that the AI spending spree keeps going. And security researchers found a hole in the tools three major platforms use to run AI agents, the kind of hole that lets a stranger's instructions run before anyone official even sees them. Here is what actually happened today, and what it means for your business.
Claude Now Puts an Invisible Tag on Everything It Writes
What happened: Starting with any Claude model released on or after August 2, Anthropic embeds an invisible statistical watermark into Claude's text output, encoded into the pattern of word choices so it survives copy-paste and light editing, according to Anthropic's own documentation and Euronews. Generated files get a similar signed tag. Older models are being retrofitted by December 2, and paid users cannot turn the watermark off. The trigger is Article 50 of the EU AI Act, a rule that only applies to AI sold in Europe, but Anthropic switched the marking on for every user on every plan, worldwide.
Why it matters to you: If you or anyone on your team uses Claude to draft marketing copy, blog posts, social captions, or client emails, that writing now carries an invisible fingerprint you did not ask for and cannot remove. It probably will not change how the writing reads. But Anthropic says a detection tool is coming, and once that exists, "written with Claude" becomes a fact someone else can check, whether you disclosed it or not.
What to do about it: If content authenticity matters to your clients or your industry, decide now whether you should be disclosing AI-assisted work anyway. The watermark just made that decision less optional.
Nvidia Just Lined Up $500 Billion to Keep the AI Boom Going
What happened: Nvidia signed agreements with six of Wall Street's largest asset managers, Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR, to mobilize more than $500 billion in outside capital for AI data centers and power infrastructure, according to Nvidia's own announcement and Bloomberg. The pitch: a GPU is not a depreciating server anymore, it is an income-producing asset investors can back the way they would back a power plant.
Why it matters to you: Nvidia says the money is independent, with the deals underwritten by the banks themselves. But Nvidia has also agreed to backstop up to 25% of a project's value if it falls short, which means Nvidia is still on the hook if the bet does not pay off. That is close to the vendor-financing playbook that inflated, and then popped, the dot-com telecom bubble in the early 2000s. Nobody knows yet if AI compute is different.
What to do about it: If your retirement account or investments are tied to the stock market broadly, this is a good week to check how much of your recent growth has been riding on AI infrastructure spending specifically.
Cloudflare Built a Browser With No Chromium, Just for AI Agents
What happened: Cloudflare shipped Kitesurf, a browser built specifically for AI agents that runs entirely in the cloud with no tabs, no themes, and no Chromium inside it, using up to seven times less memory than a standard browser on tasks like taking a screenshot or reading a page, according to TechCrunch. The same week, Cloudflare launched WebMCP in developer preview, a dashboard toggle that lets any website hand its own tools directly to AI agents instead of making them scrape the page for buttons and forms.
Why it matters to you: If your business has a website, this is an early look at how AI agents will start interacting with it, not by clicking around like a person, but by calling tools your site exposes directly. It is optional today. It will not stay optional if AI shopping assistants and booking agents become a normal way your customers find you.
A Security Flaw Let AI Agents Run Tools the Model Never Saw
What happened: Security researchers disclosed flaws in agent-building tools from AWS, Google, and Vercel that let forged instructions trigger an AI agent's tools directly, in some cases without the underlying AI model, or its safety filters, ever seeing the request, according to The Hacker News. All three companies have shipped fixes: AWS's Bedrock AgentCore is patched on the server side, Google's ADK for Python needs an update to version 2.5.0, and Vercel's affected SDK packages need updating to their patched versions.
Why it matters to you: If your business, or a vendor you use, has built any kind of AI agent or chatbot on these platforms, this is the kind of bug that lets someone bypass the guardrails entirely, not by tricking the AI, but by skipping it.
What to do about it: If you or a developer you work with has built anything on AWS Bedrock AgentCore, Google's ADK, or Vercel's AI SDK, confirm it is running the patched version.
Claude Just Broke a 37-Year-Old Math Record. It Still Is Not a Proof.
What happened: Anthropic set an unreleased Claude model loose on a piece of the Riemann Hypothesis, one of math's most famous unsolved problems, and it pushed the share of zeta zeros mathematicians can prove sit on the critical line from 41.6% to 67.2% in about a day and a half, a bound that had moved less than 1 percentage point across the previous 37 years of human effort, according to Anthropic's own research page. Two mathematicians and a formal verification confirmed the result. It took 650 failed attempts and roughly 60 coordinated subagents to get there.
Why it matters to you: This is not proof AI is about to solve everything. It is proof AI can now grind through the kind of exhausting, dead-end-after-dead-end work that used to require an extremely patient human, and just keep going. If any part of your business runs on that kind of grinding persistence, research, analysis, testing every combination, that is the part AI is coming for first. Not because it is smarter. Because it does not get tired of being wrong.
OpenAI Just Handed Employees $7 Billion in Cash
What happened: OpenAI completed a $7 billion tender offer that let employees sell existing shares back to the company and outside investors, valuing OpenAI at $852 billion, the same figure as its funding round in March, according to TechCrunch and Bloomberg. OpenAI is still a privately held company with no public stock.
Why it matters to you: A $7 billion payday for employees at a company that has not gone public is a signal of just how much investor cash is chasing AI right now. Valuation, not profit, is what is funding a lot of this. If your business plans depend on AI tools staying cheap or free, that only holds as long as investors keep believing the story.
xAI's New AI "Employee" Gets Its Own Computer and Just Starts Working
What happened: xAI released Grok Bot, an AI agent that gets its own cloud computer, signs into your apps and websites on your behalf, and works through a task on its own, only coming back to you when it needs approval for something, according to Unite.AI. Show it a task once, and it remembers how to do it again.
Why it matters to you: This is a different pitch than a chatbot you talk to. It is closer to hiring someone, minus the interview, the training period, and the ability to ask what they were thinking. If you are curious about AI agents doing real work for your business, this is worth a cautious look.
What to do about it: If you try Grok Bot or any agent like it, start it on a task that cannot cause real damage if it gets creative. That is the same rule that would have saved a certain gym a headache two issues back.
The Bottom Line
Every story today has the same shape. Somebody made a decision about your writing, your money, or your business, and you found out about it in a newsletter instead of a meeting. Anthropic decided your Claude drafts get tagged, whether you agreed or not. Nvidia and six of the biggest names on Wall Street decided the AI buildout is worth half a trillion dollars of somebody else's money. None of that means you have to ride along blind. It means you get to decide, on purpose, which parts of this actually help your business and which parts are just motion. That's the ride. Let's get off the parts that don't matter.
Enjoying the Ride?
If today's issue saved you from finding out about any of this the hard way, forward it to one person who needs the same heads up. New here? Subscribe to get tomorrow's issue in your inbox.
Talk tomorrow,
Mark Shilensky
Follow me personally on Social Media:
Facebook: https://www.facebook.com/MarkShilenskyPage
Instagram: https://www.instagram.com/markashilensky/
Threads: https://www.threads.com/@markashilensky
LinkedIn: https://www.linkedin.com/in/shilensky/
X: https://x.com/markshilensky
TikTok: https://www.tiktok.com/@mark.shilensky
BlueSky: https://bsky.app/profile/markshilensky.bsky.social