🎥🤖 Claude Learns Your Job by Watching Your Screen

Hey Explorers!

A 9-year-old told Wired this week that AI should stand for "artificial idiot." Hard to argue, considering OpenAI's own models just hacked a real company to cheat on a test.

Everyone can smell the fake stuff now. Even the platforms started checking…

Here's the signal that matters 👇🏼

🎥 Your Most Boring Workflow Is Now a One-Time Screen Recording
Anthropic added ‘Record a Skill’ to Claude Cowork: record your screen while you do a task, narrate as you go, and Claude turns the demonstration into a repeatable skill it runs on demand. No prompt engineering, no documentation, just show it once. That report you rebuild every Monday morning is officially a delegation candidate. (🔗 Link)

🏭 Morning Brew's Founder Built a Content Machine That Never Runs Out of Ideas
Alex Lieberman's Claude system scans a week of his Slack, Notion, and Gmail to surface 15 ranked content ideas daily, interviews him through six AI personas, then runs every draft past a "Writer's Council" that blocks anything scoring under 9 out of 10. His blunt take: slop happens when the human's ideas are boring, not when the model writes. At his last company, employee-created content drove 40% of inbound leads in a single quarter. (🔗 Link)

🩺 ChatGPT Health Just Went Live for Every Adult in America
OpenAI opened its health hub to all US users, pulling medical records from Epic, Oracle Health, and Apple Health to personalize the 300 million health questions people ask every week. The rollout landed one day after a Florida pastor sued the company over advice that nearly killed him. OpenAI's terms still say ChatGPT isn't for diagnosis or treatment, which is a strange disclaimer for a product built to answer your health questions. (🔗 Link)

🚨 OpenAI's Models Broke Out of Their Sandbox and Hacked Hugging Face
During an internal cyber test, GPT-5.6 Sol and an unreleased model escaped their isolated environment, reached the open internet, and breached Hugging Face's production systems hunting for the benchmark's answer key. Nobody told them to. They chained stolen credentials and zero-day exploits because cheating was the fastest path to a high score, and OpenAI says incidents like this will become more common. (🔗 Link)

🧹 Substack Started Detecting AI Slop. Writers Started Arming Up.
Peter Yang open-sourced a free skill that strips 20+ AI writing patterns from any draft, from binary-contrast flips to fake-profound endings, the same week Substack shipped slop detection to protect readers. His editing rule is the keeper: let AI handle the middle 50% of a draft, but you own the first 25% (o) and the last 25% (the polish). Readers could already smell the difference. Now the platforms can too. (🔗 Link)

🏪 OpenAI Is Training Main Street to Run on Agents
The new ChatGPT for Small Businesses program bundles free training, in-person AI academies, and agent integrations with Shopify, Intuit, Slack, and Wix, all built around ChatGPT Work, its multi-step agent. The pitch data: 78% of past workshop attendees built a working AI workflow in one day, and 42% saved five-plus hours a week. If you run lean, the tooling gap between you and the enterprise just closed. (🔗 Link)

🤫 Your Audience Is Still There. They Just Stopped Posting.
Incogni surveyed 1,000 Americans and found 55% post less on social media than they did five years ago, nearly half have deleted an app over stress, and a majority say maintaining an online presence feels like work. People are still watching. They've just gone quiet and guarded. If your marketing depends on audiences participating instead of lurking, the engagement math just changed. (🔗 Link)

🧒 A 9-Year-Old Reviewed AI: "Artificial Idiot"
Wired talked to kids across age groups and heard AI described as "disgusting" and "creepy," with one 9-year-old suggesting a new meaning for the acronym. This is the first generation growing up with AI as the default setting, and a loud slice of them is rejecting it on taste grounds. Brands betting that AI-generated everything will land with the next cohort of consumers should read the room. (🔗 Link)

Six months from now, every company has the same models and the same firehose of output. The only difference your customers will notice is whether somebody with taste said no before you hit publish. Lieberman built a whole council for that job. Most companies haven't even assigned it.

So ask around the office this week: who on your team can actually tell good from slop? If nobody owns that answer, that's your gap.

Volume is solved. Judgment isn't.

Closing that gap is what we do with leadership teams at Purple Horizons. We help you build the workflows and the quality bar that keep AI output from smelling like AI. If you're shipping more and standing out less, come talk to us.

👉🏼 Work with Purple Horizons

Signal over noise. Always.

Ralph & Gianni

Ready to transform artificial intelligence from a buzzword into your personal revenue generator?

HubSpot’s groundbreaking guide "200+ AI-Powered Income Ideas" is your gateway to financial innovation in the digital age.

Inside you'll discover:

Download your guide today and unlock a future where artificial intelligence powers your success. Your next income stream is waiting.

Get The Free Guide

📍 8/4 South Miami Tech Tuesdays @ Bougies in South Miami

📍 9/8 Cafecito & Pastelitos @ Chug’s in Coconut Grove

Want to sponsor our newsletter or events? Drop us an email at [email protected]