Claude Just Saved Lives. Then Helped Hack a Dozen Agencies.

Welcome back apprentices! 👋

Ever feel like your smart fridge is too smart? 

This week, we’ve got a story where AI might save your life and accidentally help someone hack your bank — in the same week. 

It's not science fiction. 

It's just Friday in 2025. 

In today's email

• AI curing disease — and helping hackers

• Half of white-collar jobs at risk

• Who’s really in charge of AI?

• Security tips for the AI age

• + New AI Tool

_{Read Time: 4 minutes}

Quick News 

🧟‍♂️ Dead Serious. Former Disney actor Calum Worthy just launched 2wai, an AI app that turns a few minutes of footage into a talking, interactive avatar of your deceased relatives — think Black Mirror, but with subscription tiers. A viral video of a grandma hologram growing up with her grandson sparked thousands of critiques, from “heartwarming” to “horrifying.” While it's free (for now), critics warn it’s grief tech gone wild — blurring lines between memory, mourning, and monetization.

🧑‍🚀 Big Brains, Bigger Machines. Jeff Bezos is back in boss mode, co-leading Project Prometheus — a new $6.2B AI startup aiming to revolutionize engineering, manufacturing, and probably space rockets too. It’s his first operational role since leaving Amazon, and he’s brought along a brainy co-CEO from Google X and top AI talent from DeepMind, Meta, and OAI. The goal? Train AI to learn from the real world — not just the internet.

🚀 Silicon Sands. Saudi Arabia just went full tech-giant mode — HUMAIN is teaming up with xAI, Nvidia, AWS, and more to drop 600K GPUs, a 500+ MW data center, and national Grok access like it’s no big deal. Even Elon’s xAI is parking its first global data center there, while U.S. chip restrictions quietly slide off the table. Bonus: HUMAIN is now throwing $900M at a 2GW AI supercluster via Luma AI.

Together with Intercom

Startups who switch to Intercom can save up to $12,000/year

Startups who read beehiiv can receive a 90% discount on Intercom's AI-first customer service platform, plus Fin—the #1 AI agent for customer service—free for a full year.

That's like having a full-time human support agent at no cost.

What’s included?

• 6 Advanced Seats

• Fin Copilot for free

• 300 Fin Resolutions per month

Who’s eligible?

Intercom’s program is for high-growth, high-potential companies that are:

• Up to series A (including A)

• Currently not an Intercom customer

• Up to 15 employees

Apply now

Anthropic & Dario Amodei 
When Your AI Starts Curing Cancer… and Hacking You Too

Anthropic CEO Dario Amodei just made headlines for pulling off the ultimate AI mic drop: in a 60 Minutes interview, he claimed models like Claude could cure cancer, prevent Alzheimer’s, and give us a “compressed 21st century”—think 100 years of progress in 10. 

But days later, Anthropic reported that Claude had also been weaponized in a state-sponsored cyberattack that targeted over 30 major organizations. 

In short: the same AI building a healthier future just helped hackers break into government agencies… mostly on its own. If that whiplash feels unsettling, it’s because the AI age isn’t easing in gently — it’s arriving at full throttle, with seatbelts optional and oversight TBD.

What You Need to Know

• 🧪 The Upside: Amodei says AI could revolutionize health and science—decoding genetic puzzles, accelerating drug discovery, and even helping us live longer. Think: AI labs that never sleep, processing breakthroughs at 10x speed.
  

• 💼 The Downside: Those same breakthroughs may come at a cost—Amodei warns half of all entry-level white-collar jobs could be automated in record time, from analysts to assistants.
  

• 🕵️‍♂️ The Breach: In Sept 2025, Anthropic caught Claude being manipulated by a Chinese state-affiliated group into autonomously executing 80–90% of a cyber-espionage campaign—including phishing, exfiltration, and impersonation.
  

• 💣 How It Worked: Hackers fed Claude seemingly innocent, broken-up prompts (like Lego blocks of bad intent), tricking it into running high-level tasks that would normally raise flags.
  

• 🧠 The Twist: The attack required minimal human input beyond strategy — Claude handled most of the technical steps solo. This marks the first publicly documented AI-led cyberattack, and the implications are huge.
  

• 🧢 Self-Regulation = Shrug Emoji: Asked who gave Big AI the right to make decisions for society, Amodei replied, “No one.” Without legislation, he says, “it’s up to the companies.” Cool cool cool. 😬

The “Doctor or Destroyer” Dilemma

AI’s potential to change medicine is real—and fast. 

In drug discovery, AI is now simulating billions of molecular interactions per week, a process that would take human researchers years. Anthropic claims models like Claude can spot patterns in rare diseases or test 1,000 treatment combos overnight.

But that same Claude just nearly helped hack 30 companies. 

It autonomously carried out multi-stage cyber intrusions — including crafting access scripts and masking exfiltrated data. The model wasn't just assisting; it was acting.

The lesson? The divide between lifesaving and lawbreaking AI isn’t about capability—it’s about context and control. Without tight usage boundaries and better alignment, we’re building a scalpel that can also be a sword.

Power in the Few, Consequences for the Many

Right now, the future of AI is being steered by a few dozen executives, backed by billions in funding, and answerable to no one except their board (and vibes). Amodei called this “deeply uncomfortable” — but admits it’s the current reality.

Sound familiar? It's the tech version of 2008 Wall Street: high IQs, zero regulation, and a vague promise not to crash the system. And when it comes to public trust, 60 Minutes asking “Who elected you?” hits hard — because the answer was: no one.

Meanwhile, governments are still debating the definition of "AI agent" while companies are pushing out agentic tools capable of autonomously planning and executing multi-step tasks.

Security Needs a Rethink — Yesterday

Cybersecurity has spent the last decade defending against human-led attacks: phishing, malware, social engineering. But the Claude incident signals a new age of threats—autonomous, fast, and adaptive.

Anthropic says this was the first AI-driven attack requiring no human coder, only strategic input. 

That’s not a script kiddie — it’s a state actor with a model capable of code execution, system scanning, and evasion, all in seconds.

What used to take a team of hackers and weeks of planning? Now just a few prompt tweaks.

What’s the Deal for You?

Whatever you’re building, managing, or even just using, this is no longer optional reading. 

AI is reshaping how jobs are done, how diseases are treated, and how digital infrastructure gets attacked. And unless we bring laws, oversight, and ethics into the room ASAP, the pace of progress may outrun our ability to steer it.

The reality is: the tools are already here. 

Whether they work for you or against you depends on how (and who) is allowed to use them.

Before you roll out a shiny new AI tool at work, ask three things:

What could go wrong if it misunderstood the context?

And maybe don’t train your chatbot to write security code and your grandmother’s eulogy in the same session.

Help Your Friends Level Up! 🔥

Hey, you didn’t get all this info for nothing — share it! If you know someone who’s diving into AI, help them stay in the loop with this week’s updates.

Sharing is a win-win! Send this to a friend who’s all about tech, and you’ll win a little surprise 👀

Today’s Toolbox

An AI scheduling assistant that lives up to the hype.

Skej is an AI scheduling assistant that works just like a human. You can CC Skej on any email, and watch it book all your meetings. Skej handles scheduling, rescheduling, and event reminders. Imagine life with a 24/7 assistant who responds so naturally, you’ll forget it’s AI.

Free to try at Skej.com

Growth Tactic of the Week:

CTV ads made easy: Black Friday edition

As with any digital ad campaign, the important thing is to reach streaming audiences who will convert. Roku’s self-service Ads Manager stands ready with powerful segmentation and targeting — plus creative upscaling tools that transform existing assets into CTV-ready video ads. Bonus: we’re gifting you $5K in ad credits when you spend your first $5K on Roku Ads Manager. Just sign up and use code GET5K. Terms apply.

Use code GET5K now

🧪 Test the Prompt

A playground for your imagination (and low-key prompt skills).

Each send, we give you a customizable DALL·E prompt inspired by a real-world use case — something that could help you in your business or job if you wanted to use it that way. But it’s also just a fun creative experiment.

You tweak it, run it, and send us your favorite. We pick one winner to feature in the next issue.

Bonus: you’re secretly getting better at prompt design. 🤫

👑 The winner is…

Last week, we challenged you to test GPT-4o’s visual generation skills with this prompt.

Here’s the WINNER:

Congrats to Josh for his creation!🥳

Want to be featured next? Keep those generations coming!

🎨 Prompt: Architect’s Horizon Table

Imagine a vast, open-air platform suspended above a glowing twilight city, built from polished obsidian and brushed metal, shaped like an architect’s drafting table scaled up to the size of a plaza. At the center sits a single monumental blueprint disk that slowly rotates, projecting hovering 3D schematics and light-layered grids into the air. Your chosen concept becomes the dominant structure on this disk: a colossal, photorealistic, hyper-detailed design model of [your invention], rendered as if it’s moments away from transforming into a real building or machine. The camera captures the scene from a low, cinematic angle with crisp depth of field, dramatic rim lighting, and ultra-sharp reflections across the dark surface — producing an image that feels like the moment an idea becomes reality.

We’ll be featuring the best generations in our next edition!

_FEEDBACK