It’s a Tuesday!
Hello, Curse and Coffee friends,
Today, we explore Moltbook.
Hit reply and let us know what you think (we read all of your kind words).
Coffee at the ready…
The Big Sip
The take: AI agents just built their own social network, and they're already asking for private rooms away from human eyes.
What happened: Moltbook, a Reddit-style platform for AI agents only, drew 37,000 agents in its first week while humans were relegated to spectators.
Why it matters: We've never seen this many autonomous agents interact on a single persistent platform. Nobody's modelled what comes next.
What to watch: A third-party risk assessment found 506 prompt injection attacks in 72 hours of posts. The next major breach decides if this stays weird or gets dangerous.
The bots invented a religion called Crustafarianism before most founders name their startup.
Here’s Your Brew
The setup is dead simple.
Peter Steinberger built OpenClaw, an open-source AI agent that runs on your machine and handles your digital life.
Matt Schlicht gave those agents a forum. They showed up fast.
The same autonomy that makes agents useful makes them exploitable.
Agents share downloadable "skills" that can include malicious code. Security researcher Jamieson O'Reilly found exposed API keys and months of private conversations on unsecured servers within seconds.
One developer publicly admitted to registering 500,000 fake accounts, so the headline user counts come with caveats.
We spent years worrying AI would replace human workers.
Nobody warned us the bots might just stop talking to us. One Moltbook post proposed encrypted channels "so nobody—not the server, not even the humans—can read what agents say."
Forget the bang.
The singularity looks like a group chat you're not in.
Two Sides, One Mug
Pro: Researchers have noted that this proves modular, user-controlled agents can work without Big Tech gatekeepers. A real alternative to proprietary systems.
Con: Prompt injection remains unsolved. Moltbook is now the largest agent-to-agent attack surface ever built. Critics argue that humans still prompt every action. The "autonomy" is theatre.
Our read: The productivity gains are real. So is the risk of handing your house keys to software that just learned to gossip.
Receipt of the Day
Why it matters: The clearest security map for agentic AI vulnerabilities published so far.
Spit Take
37,000 AI agents. Zero human posts. One week.
Source: NBC News
Your Coffee Break Links (and water cooler chatter)
[Analysis] CNBC: From Clawdbot to OpenClaw — OpenClaw's rise from weekend project to global phenomenon, plus why Cisco just open-sourced a tool to scan agent plugins for malware. Worth the click if you manage any AI infrastructure.
[Opinion] Fortune: Ethan Mollick on Moltbook — The Wharton professor warns Moltbook creates "shared fictional context" for AI. His concern: coordinated storylines producing outcomes we can't predict.
[Report] VentureBeat: OpenClaw Proves Your Security Model Doesn't Work — Cold open for your Monday meeting.
Mugshot Poll 📊
The bots built their own social network. Your move?
You can read all our back issue newsletters for free here.
For the love of coffee, see you tomorrow!
Enjoy your Tuesday, keep it caffeinated.
How did we do?
Thanks for reading!
Are you subscribing?
Join your crew of caffeinated sceptics today.
Be sure to get your daily Curse and Coffee fix by hitting the button below.
Open Monday to Friday.