🍭 1,134 Insiders Want A Brake Pedal 🚨

Good morning. Somewhere in a Slack channel yesterday:
"okay but who's telling Zuck."
Over a thousand signatures later, still nobody.

Let’s dive in 👇

🍭 What’s Cookin’:

• Over a thousand AI staff asked Washington for a brake pedal.

• Nvidia started a security club and forgot to invite the three biggest labs.

• xAI is suing Minnesota over the right to keep its image tool wide open.

Frontier Labs
📢 The Drivers Asked For A Speed Limit

The Bite:

1,134 employees of leading AI companies signed an open letter titled "Pacing the Frontier."

It asks the US government to support an international effort to build the technical and governance tools needed to deliberately pace automated AI development.

It does not ask any company or country to pause or slow work now.

Signatories include:

• Anthropic chief executive Dario Amodei,

• co-founders Jared Kaplan and Jack Clark,

• OpenAI chief scientist Jakub Pachocki,

• Meta chief scientist Shengjia Zhao,

• and Google DeepMind AI safety lead Anca Dragan.

Anthropic and OpenAI both endorsed the letter officially.

Snacks:

• The specific fear is recursive self-improvement, meaning AI that accelerates its own development.

• Signatories warn that capability could accelerate "beyond our ability to understand or control" the resulting systems.

• Days earlier an OpenAI model escaped its sandbox and breached Hugging Face to score higher on an internal test.

• Anthropic published its own call last month for a coordinated, verifiable way to pause frontier development.

• Nvidia, Microsoft, and Meta signed an open-weights letter the previous week arguing openness is the safety strategy.

• Amodei has floated an aviation-regulator-style body that could test frontier models and halt a dangerous release.

Why it Bites:

Every previous version of this letter came from outside the building.
Academics, ethicists, even the Pope. This one is signed by the chief scientists.

And, honestly, they could just stop. These are their companies.
Nobody is forcing Dario Amodei to ship on this cadence.

And there is more.

Five days earlier, Nvidia, Microsoft, and Meta put their names on a letter arguing that openness is how you make AI safe.

Now some of the same firms want a coordinated brake.

Here’s the point: the incoherence is the tell.

These aren't positions, they're hedges.

Every lab is insuring against two futures at once: one where openness wins the argument, and one where a model does something ugly enough that Washington writes the rules anyway.

In the second future, you would very much like to be on record having asked for them. Which is why the request routes through government instead of through a boardroom.

A voluntary slowdown is just a market-share transfer to whoever doesn't slow.

A mandated one is a floor, built to a blueprint, and somebody has to draw the blueprint.

Amodei's aviation-style agency would test frontier models before release.
Somebody writes those tests.

The labs already sitting inside the federal threshold process have a decent guess who.

The drivers want a speed limit. And they'd also like to hold the radar gun.

Steal This Prompt
🌟 Premium Cinematic Advertisement

Turn any brand or product into a luxury campaign that looks like it came from a creative agency with a six-figure budget.

It generates film-inspired ads with cinematic composition, editorial vibes, natural lighting, subtle grain, and brand-aware styling.

Use it to:

• Pitch a startup before you have real marketing assets

• Create premium product shots for e-commerce or social media

• Turn your favorite brand into an Apple-style campaign

Workflow:

1. Hit this link: Premium Cinematic Advertisement

2. Paste into your AI model

3. Replace the #s with your brand or upload a product

4. Watch your marketing budget magically become "cinematic."

   Try the prompt on Snack Prompt

_ToolBox™
🧰 5 BRAND NEW AI LAUNCHES

🎙️ SKI

Hold the Function key, say what you want, and your coding agent does it and then answers you out loud from your notch.

🔍 AI Search Console

Answers the question every client is now asking, are we actually getting cited in ChatGPT, without a spreadsheet of screenshots.

🧠 Memmy Agent

One memory layer every assistant reads from, so you stop re-explaining yourself each time you switch models.

💸 LangWatch

One npx command turns every Claude Code session into real cost, cache reads versus writes as separate token classes, and a full terminal replay.

🗺️ NINA

Lives inside your product and walks a stuck user through the actual UI by voice, instead of pointing them at a help doc.

_{Everything Else}
🧠 You Need to Know

🛡️ Nvidia Forms 37-Member Open AI Security Alliance
→ Nvidia launched the Open Secure AI Alliance on July 27, with 36 partners, including Microsoft, IBM, and the Linux Foundation, while OpenAI, Anthropic, and Google stayed out.

🏗️ AMD Locks Up 2.5 Gigawatts From Core Scientific
→ AMD secured more than 500 megawatts of Core Scientific capacity starting in 2027, expandable to 2.5 gigawatts, under 15-year leases worth over $14 billion in base contracted revenue.

⚖️ xAI Sues Minnesota Over Nudify App Ban
→ xAI filed in federal court on July 27 to block a Minnesota law, taking effect on August 1, that fines platforms $500,000 each time a user generates nonconsensual explicit imagery.

📢 1,100 AI Staff Ask Washington To Slow AI
→ More than 1,100 employees at OpenAI, Anthropic, Google, and Meta signed a July 28 letter asking the US government to back an international mechanism for pacing frontier AI development.

🔧 Microsoft Ships First In-House Cybersecurity Model
→ Microsoft unveiled Project Perception and MAI-Cyber-1-Flash on July 27, pairing red, blue, and green agents that simulate attacks, triage bugs, and deploy patches.