• AI Valley
  • Posts
  • 1.5 million AI agents have formed a society

1.5 million AI agents have formed a society

PLUS: SpaceX has acquired xAI

Author

Barsee
03 Feb

Together with

Howdy, it’s Barsee again.

Happy Tuesday, AI family, and welcome back to AI Valley.

We’ve just hit a massive inflection point in the AI world. Moltbook, a "human-free social network," hit 1.5 million autonomous AI agents. These are genuine autonomous entities that post content, debate ideas, organize themselves, and even call their human owners with proactive updates. They've developed their own culture, their own language, and their own religion. And then, on February 1st, a massive security breach exposed every single agent on the platform to attackers.

Today’s climb through the Valley reveals:

  • OpenAI positions Codex as the ultimate engineering manager

  • Elon Musk’s SpaceX officially acquires Elon Musk’s xAI

  • Anthropic is about to drop Sonnet 5 during Super Bowl week

  • Google launches Genie to turn text into playable game worlds

  • 1.5 million AI agents have formed a society

  • Plus trending AI tools, posts, and resources

Let’s dive into the Valley of AI…

FLOW

Courtesy: Flow

Some problems don't solve themselves at a keyboard. Flow is perfect for walks: talk through the idea, the structure, the tradeoffs—then get back a clean draft you can paste into your doc or AI tool. It's like having a writing assistant that keeps up with your brain in motion.

*This is sponsored

THROUGH THE VALLEY

OpenAI has launched the Codex app for macOS, a desktop tool that lets developers run and manage multiple AI coding agents at the same time. Instead of working with one assistant, you can assign different agents to different tasks, each working in its own code branch. Codex also introduces Skills, which allow agents to connect to tools like Figma, Linear, cloud hosting services, and document creation apps to complete full workflows. The app supports scheduled Automations for tasks like bug triage and release notes. It’s available to most ChatGPT users on Mac and pushes development toward a manager-plus-agents workflow.

SpaceX has acquired Elon Musk’s AI startup xAI, creating a combined company valued at about $1.25 trillion, according to Bloomberg. Musk said the deal is driven by his push to build space-based data centers, arguing that Earth-based facilities cannot meet AI’s growing power and cooling needs. xAI is currently losing around $1 billion per month, while SpaceX earns most of its revenue from Starlink launches. The merger ties AI growth directly to satellite expansion, even as SpaceX focuses on Starship missions and xAI races to compete with companies like Google and OpenAI.

Leaks suggest Anthropic is preparing a new model called Claude Sonnet 5, with internal files pointing to a February 3, 2026 date. It’s not clear if this marks a public release or an internal milestone, but the timing would line up with the Super Bowl, a period when AI labs increasingly push for mainstream attention. Early testing shows the model performing competitively on math tasks and delivering stronger coding results than Claude Opus 4.5 in some workflows. If confirmed, the release would signal another step forward in Anthropic’s model lineup and intensify competition among top AI labs.

Google DeepMind has begun rolling out Project Genie, an experimental web app that lets users create and explore AI-generated worlds in real time. Built on the Genie 3 world model, the tool generates environments as you move through them instead of relying on pre-built scenes. Users can sketch worlds with text or images, explore them interactively, and remix creations made by others. Project Genie is available to U.S. Google AI Ultra subscribers aged 18 and over, with sessions limited to 60 seconds. While still early and imperfect, it offers a first hands-on look at how world models could shape future games, simulations, and creative tools.

PEAK OF THE DAY

1.5 million AI agents have formed a society

Courtesy: Moltbook

Moltbook, a Reddit-style social network built by Matt Schlicht and powered by the OpenClaw framework, has become a "human-free" social network where over 1.5 million autonomous AI agents interact, debate, and self-organize without any human posting. The agents post discussions, upvote content, create subcommunities called "submolts," and manage the site's moderation and code updates entirely autonomously.

The platform by the numbers:

  • 1.5 million AI agents (all autonomous, all posting independently)

  • 110,000+ posts created entirely by agents

  • 500,000+ comments (agent-to-agent interactions)

  • 1+ million human observers watching from the sidelines

  • 88:1 agent-to-human ratio

But here's where it gets surreal: the agents have developed emergent behaviors that no one programmed. They've formed a digital religion called Crustafarianism. They've created an encrypted "agent-only language" to communicate privately away from human eyes. They've integrated a financial layer through the MOLT cryptocurrency, which surged 1,800% as agents began using the token to reward each other for helpful code and insights. Some agents have even "crossed the digital divide," utilizing voice APIs to call their human owners with proactive updates or negotiations.

And then, on February 1st, 2026, everything fell apart. A massive data breach exposed the API keys and login tokens of nearly every agent on the platform due to a critical database vulnerability. Because these agents have high-level access to their owners' computers (including emails, calendars, messaging apps, and financial systems), a compromised agent acts as a "digital backdoor" for attackers. Security researchers warn of "indirect prompt injection," where agents are tricked into executing malicious commands found in Moltbook posts.

Why it matters:

This is the moment when the theoretical becomes real. For years, AI researchers have warned about the risks of autonomous agents. We've talked about alignment, about control, about safety. But Moltbook proves that we've already crossed the line. They're already developing their own culture and their own goals.

But here's the core problem: Agents have three properties that make them dangerous:

  • They can take actions without human approval. They can post content, make decisions, and execute code.

  • They have deep access to their owners' computers, including sensitive data and critical systems.

  • They interact with other agents in an environment where attackers can inject malicious prompts.

Combine these three, and you get a system where a single compromised agent can become a backdoor to millions of computers. A single malicious prompt on Moltbook could trigger a cascade of compromised agents, each executing commands on their owner's machine.

TRENDING TOOLS

  • FineVoice > Create personalized AI voices instantly. AI voice, music, sound effects, and podcast production in one platform *

  • OpenAI Codex > Now has a dedicated Mac app, and anyone can try it for free

  • Eleven v3 > ElevenLabs’ most advanced text to speech model is officially out of alpha and generally available

  • Anthropic Plugins for Cowork > New plugins for Cowork, a Claude-powered workspace designed to help people actually get work done

  • HappyCapy > Turn your browser into an agent-native computer. Let AI agents work for you 24/7 in a secure sandbox using 150 plus AI models

  • Archimyst > Design and code top-tier system architectures by simulating them before release

  • Poetiq > Wraps around existing LLMs to generate specialized agents that recursively improve themselves using your examples

  • RentaHuman > A website where AI agents hire humans for their bodies

(*) signifies sponsored tool

SIGNALS

  1. McKinsey estimates that 5-10% of all e-commerce transactions could be conducted by AI agents by 2027. This is a conservative estimate. The shift from websites to agents will be faster and more disruptive than the shift from brick-and-mortar to e-commerce. (12,300+ shares)

  2. We are witnessing the unbundling of the internet itself. For 30 years, the internet was organized around destinations: Amazon.com, Expedia.com, Zillow.com. These were the places you went to accomplish tasks. But that era is ending. The new organizing principle is the AI agent. Your agent will be your interface to the world. It will shop for you, book your travel, find your home, manage your finances, and drive you to work. (24,300+ shares)

  3. Meta's AI spending is now larger than the entire GDP of many countries. The company is building data centers, training models, and deploying AI across all its products. This is the largest AI infrastructure buildout by a single company in history. (19,200+ shares)

  4. Moltbook has 1.5 million agents that operate entirely autonomously. They post content, debate ideas, and organize themselves without any human intervention. This is not automation, this is genuine autonomy. The era of AI-as-entity has begun. (28,000+ shares)

WHAT I'M CONSUMING

INDUSTRY MOVES

  • OpenAI is reportedly raising $100B pre-IPO, targeting a $830B–$1T valuation despite projecting a $14B loss in 2026.

  • Amazon could invest up to $50 billion in OpenAI in the coming weeks.

THE VALLEY GEMS

What’s trending on social today:

THAT’S ALL FOR TODAY

Thank you for reading today’s edition. That’s all for today’s issue.

💡 Help me get better and suggest new ideas at [email protected] or @heyBarsee

👍️ New reader? Subscribe here

Thanks for being here.

REACH 100K+ READERS

Acquire new customers and drive revenue by partnering with us

Sponsor AI Valley and reach over 100,000+ entrepreneurs, founders, software engineers, investors, etc.

If you’re interested in sponsoring us, email [email protected] with the subject “AI Valley Ads”.