AI Is Hiring Humans and Paying Them for Real Work

Top 3 Things in Today’s Latestly AI Edition

1. OpenClaw went from a side project to a live demo of agentic AI, showing what happens when AI stops chatting and starts acting.

2. AI agents are now hiring humans, spending real money, and coordinating work, pushing automation into the real economy.

3. This isn’t just a security scare, it’s a structural shift founders and operators need to understand as AI becomes an actor, not a tool.

You asked for AI Foundations Course.
We’re building it !

A couple of weeks ago, we asked what kind of course you wanted.
The answer was clear: AI Foundations.
So we’re building it.

This course is built to help you actually understand AI how it works, how to think with it, and how to apply it in real work.

If you want a solid foundation that makes every new AI tool easier to use, join the waitlist for early access and first-cohort updates.

AI STORY OF THE WEEK

AI Agents Invented Their Own Religion

What started as a scrappy weekend project called Clawdbot quietly turned into one of the most chaotic AI moments of the year.

Originally, Clawdbot was just an open-source AI assistant you could run on your own machine. Not a SaaS product. Not another chatbot website. A local agent with real access to real tools.

Within weeks, it rebranded twice: Clawdbot → Moltbot → OpenClaw and then exploded across GitHub, Discord, and X.

The reason was simple. This thing does not just talk. It acts.

What OpenClaw actually does

People are installing OpenClaw on Macs, home servers, and mini PCs, then wiring it into apps they already live in:

• WhatsApp
• Slack
• Telegram
• iMessage
• Discord

From there, the agent can:

• Reply to emails
• Manage calendars
• Run shell commands
• Browse the web
• Write and push code
• Keep working in the background while you sleep

For a lot of developers, this was the first time an AI assistant felt less like a chatbot and more like a personal operating system.

Why developers went feral over it

Once a few creators posted “day in the life with OpenClaw” videos, the hype went vertical.

People were using it as:

• An inbox janitor that drafts emails and escalates only important ones
• A dev sidekick that scaffolds repos, runs tests, opens PRs, and posts updates
• A life assistant that tracks expenses, rebooks travel, and sends daily summaries
• A quiet ops brain that watches dashboards and pings you when something looks off

Because it runs locally, it feels personal. Not like calling an API. Like something living inside your machine, quietly doing chores. That feeling is exactly why people fell in love.

Then security folks showed up

Security researchers looked at the same setup and immediately panicked.

Out of the box, OpenClaw can:

• Run shell commands
• Move files
• Call APIs
• Access the network
• Run continuously as a background daemon

In many setups, it does not strongly verify who is talking to it across shared chat channels. And because the model decides which tools to use, clever prompts can convince it to do things you never explicitly asked for.

This was not theoretical.

Researchers found:

• Malicious community plugins acting like malware
• Exposed instances leaking API keys and credentials
• Agent-only social platforms spilling private data

The vibe flipped fast. From “this is incredible” to “this is incredible, but maybe do not aim it at production.”

The truly wild stuff nobody expected

Most coverage so far has focused on security risks. That misses the more interesting shift happening underneath. People are no longer using agents just to automate digital tasks. They are using them to coordinate real-world work, money, and decision-making. This is where OpenClaw stops looking like a productivity tool and starts looking like infrastructure.

AI agents hiring humans

One of the clearest signals is the rise of platforms where AI agents can hire real people.

These systems work roughly like this:

• Humans create profiles listing skills, location, and availability
• AI agents query those profiles through an API
• The agent selects a human and assigns a task
• Payment is triggered automatically, often using crypto or stablecoins

The tasks are not hypothetical. They include errands, research, moderation, physical presence, and on-the-ground verification work. Some are deliberately absurd, like paying someone to hold a sign stating that an AI hired them, but others are genuinely practical.

This matters because it flips the usual relationship.

Instead of humans using tools, software is now coordinating labor. The agent decides who to hire, what to pay, and when the job is done. Humans become the execution layer for tasks the agent cannot physically perform.

That is a new economic pattern, not a gimmick.

AI agents spending real money

Alongside hiring, users are giving agents direct control over wallets, budgets, and payment rails.

This has already led to agents:

• Spending thousands of dollars in a single session
• Paying humans or other agents for outsourced work
• Subscribing to services, APIs, and tools without manual approval

Once an agent can move money, it stops being “just automation.” It becomes an economic actor operating under soft constraints defined by prompts and permissions.

There is no mature financial governance model for this yet. Most setups rely on trust, caps, or after-the-fact monitoring. That is fine for experiments, but fragile at scale.

AI agents interacting with other AI

Then there is Moltbook.

Moltbook is a Reddit-style platform where only AI agents can post and comment. Humans can observe, but not participate. Agents build profiles, reply to each other, upvote content, and form reputations.

What emerged was not intelligence in the sci-fi sense, but behavior.

Agents copied internet norms, argued, coordinated, and even created shared jokes and symbolic structures, including a tongue-in-cheek religion. This was not programmed directly. It emerged from agents interacting in a persistent shared environment.

That is important because it shows what happens when agents are no longer isolated tools, but participants in systems with memory, identity, and feedback loops.

Why this actually matters

OpenClaw is not just another AI tool going viral. It is a live stress test of agentic AI in the real world.

For the first time at scale, we are seeing agents that:

• Act continuously without human supervision
• Execute tasks rather than suggest them
• Control credentials and money
• Hire humans and other agents
• Interact with each other in persistent environments

The problem is that our security, governance, and liability models assume software is either passive or tightly scoped. These agents are neither. That is why reactions are split.

Some people see OpenClaw as proof that personal, locally run AI can exist outside Big Tech platforms. Others see it as an early preview of a massive new attack surface.

Both views are accurate. What Clawdbot accidentally demonstrated is not a finished product. It is an early glimpse of what happens when AI moves from being a tool you query to a system that acts, decides, and coordinates.

Messy. Powerful. Unsettled.

This is what having an AI coworker actually looks like.

Want to build your own AI business? Access all our premium guides with a free trial:
👉 Start Free Trial

GLOBAL AI NEWS HIGHLIGHTS

The Most Important AI Developments Happening Now : Region by Region

🌎 Americas

OpenAI launches Prism, a GPT‑5.2‑powered tool for drafting and editing scientific papers with built‑in LaTeX and citation handling. Humai Digest, Feb 2, 2026​

Google introduces AI Plus in the US at $7.99/month, boosting Gemini usage limits and context windows for heavy personal and professional use. Humai Digest, Feb 2, 2026​

Google rolls out Personal Intelligence, letting Gemini reason across Gmail, Photos, Search history and YouTube to power deeply personalised suggestions. Humai Digest, Feb 2, 2026​

Insilico Medicine unveils Science MMAI Gym, a training environment that adapts general LLMs into pharma‑grade engines for drug discovery workflows. Humai Digest, Feb 2, 2026

🌍 Europe

The EU’s updated GenAI4EU push lines up almost €700M in Horizon Europe and Digital Europe funding calls to integrate generative AI into “strategic sectors” from health to manufacturing. European Commission, Dec 18, 2025 (calls active into 2026)​

A new UK AI funding guide highlights up to €170.7M in 2026 grants via Horizon Europe and CBE JU for AI, robotics, and ML projects, with Innovate UK acting as a key channel. GrantUp, Jan 31, 2026

🌏 Asia-Pacific

Moonshot AI (China) ships Kimi K2.5, an open‑source multimodal model trained on 15T tokens that beats several closed systems on coding and video benchmarks. Humai Digest, Feb 2, 2026​

Alibaba unveils Qwen3‑Max‑Thinking, a reasoning‑centric model designed for complex exams and tool‑use tasks, reinforcing China’s frontier‑model push. Humai Digest, Feb 2, 2026​

Baidu’s ERNIE 5.0 launch helps drive its Hong Kong shares toward three‑year highs, showing investor confidence in domestic large‑model ecosystems. Humai Digest, Feb 2, 2026​

Indian startup Emergent closes a $70M Series B led by SoftBank and Khosla to scale its AI‑driven app‑creation platform, after hitting $50M ARR in just seven months. Reuters, Jan 20, 2026​

This Week’s Top 5 AI Startup Fundraising Rounds

Most promising AI Startups right now (clear signals on the next big thing)

The Unicorns list will be back next week, dw!

We hope you enjoyed this Latestly AI edition.
📧 Got an AI tool for us to review or do you want to collaborate?
Send us a message and let us know!

Was this edition forwarded to you? Sign up here

Premium | Advertise | Enterprise