Welcome, humans.

While everyone’s been panicking about AI's insane costs, ChatGPT finally crept up on 1 billion weekly users, a milestone OpenAI wanted seven months ago. Late, yes, but still faster than TikTok, Instagram, or YouTube reached the same mark without a bumpy GPT-5 launch.

Forget the spreadsheets. A billion people now open ChatGPT weekly like they’d open Google, so “is this AI stuff actually catching on” stopped being a real question.

My group chat replaced “let me Google it” with “let me ChatGPT it” months ago, and so has basically everyone else’s.

Here’s what happened in AI today:

😸 Zuckerberg wants the U.S. to hit the AI accelerator.
📰 Wall Street is asking who blinks first on AI spending.
📰 Encore AI raised $30M to build sales and support agents.
🍪 Browser agents are moving into ordinary work apps.
🎓 Use a browser agent without letting it freelance.

...and a whole lot more that you can read about by clicking here.

P.S: We’re going live later today @ 10am PT to talk about how to get your AI app on the App Store (a continuation of last week’s live); click here to join / watch later.

😺 Zuckerberg Just Publicly Broke Ranks With OpenAI and Anthropic On Slowing Down AI

Yesterday we told you that Sam Altman and Dario Amodei backed a petition asking Washington to build the tools to "pace" AI development. Well, Mark Zuckerberg took out a Wall Street Journal op-ed to go out on his own and say: hard pass.

Here's what happened:

Zuckerberg argued the U.S. "should accelerate AI development, not restrict it," and said the real risk isn't AI moving too fast, it's "superintelligence" (AI that outperforms humans at most tasks) ending up locked inside a handful of labs.
He said distributing powerful AI broadly would create more jobs and make it easier to start a business without raising huge amounts of money, not fewer jobs like critics fear.
He also pushed back on banning Chinese open-weight AI models (models anyone can download and modify for free), calling it unlikely to actually work.
The op-ed landed the same week OpenAI and Anthropic officially endorsed a petition from 1,224 frontier-lab employees asking governments to build "the technical and governance tools" to slow AI down if it starts improving itself faster than anyone can supervise.

Why this matters: This is the clearest public split yet between two camps forming across Silicon Valley. Meta, NVIDIA, Microsoft, and Elon Musk have been championing open, fast-moving AI. OpenAI and Anthropic, the same companies racing hardest to build the most powerful models, are now the ones publicly asking regulators to install brakes. If you work anywhere AI touches your job (so, everywhere), this argument over speed vs. caution is about to shape the actual rules you operate under, not just the headlines you scroll past.

Our take: Here's the detail nobody's talking about: Meta's own chief AI scientist, Shengjia Zhao, personally signed the pacing petition, the same one his boss just spent an entire op-ed arguing against. Most corporate disagreements happen between rival CEOs. This one's happening under the same roof. Worth asking: when the people closest to the frontier keep hedging their own bosses' bets, is that a coincidence, or a signal? Anyway, Grant got around to publishing his full thoughts on this. Check out his hot take on to pace or not to pace here.

FROM OUR PARTNERS

Everyone's talking about Claude. Almost nobody is using it well.

Claude has shipped more major features this year than any AI tool: Skills, Connectors, Cowork, and vibe coding. But most people still only use it like a chatbot.

That's why we built the 2-Day Claude AI Mastery Workshop. We condensed 800+ hours of research, testing, and real-world workflows into a live, hands-on, 16-hour course.

You'll learn how to:

Master Claude's Chat, Cowork, and Code modes
Build automations with Skills, Connectors, and your favorite apps
Vibe code apps and dashboards without writing code

Saturday & Sunday • 10 AM to 7 PM ET

Register now and learn the AI skills that will save you hours every week.

🎓 AI Skill of the Day: Supervise Your Browser Agent

Browser agents are getting useful, which means they are also getting just dangerous enough to need adult supervision. Perplexity’s Personal Computer now works on Windows, where it can use local files, Microsoft 365, and the web; Polar is pitching saved prompts and tab-aware agents for knowledge work; and Google is letting Mac users call Gemini from any window with a long fn-key press.

The skill: treat a computer-use agent like an intern with temporary access, not a wizard with your house keys.

Give it one bounded job: research, summarize, draft, compare, or organize.
Name the apps or folders it may use.
Require approval before sending, deleting, buying, submitting, or changing files.
Ask for an action log so you can audit what happened.

Act as my supervised browser/computer-use assistant.
Goal: [describe the task]
Allowed sources/apps: [tabs, files, websites, or apps]
Do not send, delete, submit, purchase, install, or change files without asking first.
When you finish, give me:
1. What you checked
2. What you changed or drafted
3. What still needs human judgment
4. Any action that requires my approval

Have a specific skill you want to learn? Request it here.

🍪 Treats to Try

Polar helps you automate knowledge-work tasks from open tabs, saved prompts, and scheduled browser workflows - free with limited credits, then $20/mo.
Perplexity Personal Computer for Windows lets Max and Enterprise Max users run an agent across local files, Microsoft 365, and the web - starts at $200/mo.
Grok Build Mode lets SuperGrok Heavy subscribers prompt Grok to create and publish websites, apps, games, and dashboards - included in SuperGrok Heavy beta access.
Pangram 4 adds updated AI-text detection and a new image detector for schools, publishers, platforms, and writers - pricing not disclosed.
Replit Design turns a prompt, URL, Figma file, or screenshot into production-ready landing pages, prototypes, posters, and emails —pricing not public.
Gumloop lets your team build shared agents that automate tasks across Slack, Gmail, Salesforce, and other work apps while IT controls access, models, and spending (raised $50M) —14-day free trial, then $37/mo.
Actively assigns every sales account a 24/7 agent that researches buyers, flags pipeline risks, drafts outreach, and recommends the next move —pricing by demo.

New from The Neuron: AI Explained

In our latest episode, AWS’s Deap Ubhi explains how Startup Advisor gives you architecture, security, cost, deployment, and scaling guidance inside your coding workflow. If you’re like Grant and use Claude Code, you can install the AWS Startup Advisor plugin (just one of many options) to help you navigate AWS, deploy your apps, and turn quick prototypes into reliable software that customers can trust and pay for. 🎧 Watch / Listen here: YouTube | Spotify | Apple Podcasts

📰 Around the Horn

Marion Lepert built OpenDerm, an open-source home robot that captures high-resolution skin images and reconstructs reproducible 3D maps so people can track lesions for early signs of skin cancer (GitHub).

Big Tech’s AI spending race faced a new earnings-season test as investors pressed Microsoft, Meta, Alphabet, and Tesla to justify rising capital costs and weaker free cash flow.
Encore AI raised $30M to build sales and support agents that learn playbooks from customer calls, messages, and CRM data.
CTGT research found censorship behavior in Chinese open models may not carry over after distillation, complicating U.S. arguments for blocking Chinese model use.
Major music labels asked global chart operators to bar AI-generated songs unless the work is primarily human-made and lawful.
xAI sued Minnesota over its ban on nonconsensual intimate-image generation apps, turning synthetic-media safety into a First Amendment fight.
OpenAI reported that GPT-5.6 Sol rewrote production GPU kernels to cut serving costs 20%, while retained reasoning and compaction tripled its ARC-AGI-3 score using six times fewer output tokens.

FROM OUR PARTNERS

Own AI deployment, grow your career

Making AI actually work day to day is becoming its own job. Hear from three people doing it: Simone Santiago Broad (Yoco), Yelva Espinoza (Zumba Fitness), and Fin's Dave Lynch. They share what the role really looks like, how it came to exist, the skills worth hiring for, and the challenges they're tackling right now. Watch the full conversation on demand.

Watch now