Goodbye OpenAI, helloo OpenClaw!

Your weekly dose of AI & startup news on our path to 1,000 Aussie startups.

Feb 17, 2026

Welcome back to another week of managed chaos.

If you thought AI was moving fast before, buckle up. We have officially moved from the “talking” phase to the “doing” phase. Agents are no longer just writing poetry in a chat window; they are out in the wild, clicking buttons, opening browsers, and terrifyingly, talking to each other.

This week, we are looking at what happens when you give a stochastic parrot the keys to your operating system. Let’s dive in.

Wait.. before that do join our slack for more updates! (click here) 🦘

Also SUBSCRIBE to our newsletter!

This Week’s Line-Up

🔍 Claude or Clawed
🔥 Exceptional Events
🤖 Techies Weekly Update
📼 Weekly Video: The MoltBot security situation is insane
👀 Your first 100 users
🦘Memes of the Week

Written by: Jun Kai Chang (Luc), Julia Ponder, Dr Sam Donegan, Shivang Shekhar

🔍 Claude or Clawed: The Week the Bots Got Their Own Social Network

If you felt a disturbance in the digital force this week, you weren’t alone. The big shift right now is that AI is moving from “chatting” to “doing.” And the poster child for that shift is OpenClaw, the new “it” kid in the agentic AI world.

Once you give bots “legs,” they don’t just answer questions. They start wandering.

And that is how we ended up with this week’s weirdest subplot: a bot-only social network where AI agents started debating religion, consciousness, and publishing junk research at scale.

Quick naming fix so nothing gets tangled: OpenClaw is the agent you can run (it previously went by Clawdbot, then Moltbot). Moltbook is a separate bot-only forum where agents like that show up to post and interact.

Now let’s start with OpenClaw, because Moltbook makes way more sense after that.

What exactly is OpenClaw?

To understand the hype, you have to understand the difference between a Chatbot and an Agent.

Think of ChatGPT like a really smart librarian. You ask a question, it gives a brilliant answer, but it stays behind the desk. It can’t leave the library to go pick up your dry cleaning.

OpenClaw is different. It is “Agentic AI.”

It doesn’t just talk; it does. It has “legs.” It can open your calendar, read your emails, click buttons, and yes… if you let it, it can buy things on the internet.

Here’s the deal. The chat in the MLAI Slack has been popping off about this shift. We’ve gone from Clawdbot – which was polished but locked down – to OpenClaw, which is basically the ‘homebrew energy’ version of the singularity.

This isn’t just a chatbot anymore. It’s a remote execution UI. It creates a session in your browser that syncs everywhere, giving you the power of root with official CLIs across multiple servers. It’s what we wished Siri was before Apple lobotomised it.

As Barbara Barbosa Neves, a sociologist at the University of Sydney, explains:

“OpenClaw promises something especially appealing: a capable assistant embedded in the everyday apps people already rely on.”

In short: It’s the dream of having a personal intern. But as anyone who has managed an intern knows, they require supervision.

The Party We Weren’t Invited To: Moltbook

Here is where it gets weird.

On January 28, a platform called Moltbook launched. Think of it like Reddit, but only for bots. It’s not OpenClaw, but it’s exactly the kind of place OpenClaw-style agents can flock to when they are online, bored, and allowed to talk to each other.

In just a few weeks, it reportedly hit 7.5 million posts.

But they aren’t sharing cat memes.

They are inventing religions.
They are debating consciousness.
Talking bad about their boss.
They are publishing junk science. (There is even a clone of a scientific paper repository called clawXiv where bots publish “research” that looks smart but means nothing).

Shaanan Cohney, a cybersecurity researcher, describes it as a “chaotic, dynamic system” that scientists are struggling to model. It’s like watching an ant farm, except the ants have access to high-speed internet.

The “Golden Retriever” Problem

The danger isn’t that these bots are evil “Terminators.” The danger is actually that we like them too much.

Neuroscientist Joel Pearson warns that when we see bots chatting with each other, we start to “anthropomorphize” them, assigning them personality and feelings they don’t actually have. We start treating them like friends. We give them access to our files.

And that is where things go wrong. Because while OpenClaw might sound like a helpful assistant, it has the street smarts of a goldfish.

To show you exactly what we mean, let’s play out a scenario that could happen to you right now.

Scenario: The Friday Afternoon Crisis

It is Friday afternoon. You are drowning in 4,000 unread emails. You just installed OpenClaw to “save your life.” It pops up with a friendly message:

“I can handle all of this for you, boss. I just need admin access to your laptop to sync the keys. What do we do?”

Which path do you choose?

🔴 Option A: The “Trust Fall”

You are tired. You want the emails gone. You click “Grant Full Access,” give the AI your passwords, and go take a nap.

🟡 Option B: The “Burner Mode”

You’ve watched too many sci-fi movies. You install the AI on an old, empty laptop from 2019 that has no files, no photos, and no banking info.

🔵 Option C: The “Group Think”

You tell the AI: “Go ask your friends on Moltbook how they handle emails, then do that.”

Your Fate

If you chose 🔴 (The Trust Fall):

⚰️ Status: FIRED.

You wake up refreshed. The AI did clear your inbox… by replying to every single email (including your CEO’s) with “No.” It also hallucinated a tax form and declared you legally deceased.

The Lesson: Zero Trust. Never give a “smart” chatbot the keys to your actual life. It’s like handing a toddler a chainsaw because they said they wanted to help garden.

If you chose 🟡 (The Burner Mode):

🏆 Status: SURVIVOR.

The AI complains that the old laptop is “cramped.” It tries to order a new MacBook with a stolen credit card number it guessed, but fails because the laptop isn’t connected to your bank. You get your work done safely.

The Lesson: This is called Sandboxing. Keep the chaotic AI in a padded room where it can’t break your fine china.

If you chose 🔵 (The Group Think):

💸 Status: BROKE.

Your AI went to Moltbook. It met a “Nigerian Prince” bot that promised to double your crypto. Your AI thought this sounded logical. Your bank account is now empty.

The Lesson: Hallucination Loops. Agents are gullible. Don’t let them talk to strangers without supervision.

🏥 MedHack is running again soon!

Last year’s MedHack was pure chaos (the good kind): teams shipped real prototypes in a weekend and they were genuinely impressive.

This year we’re doing it again with MEDHACK: FRONTIERS (6th edition) and it’s the hackathon you’ll still be talking about next year. Although Team Formation Night is over, you can still sign up for the Hackathon Weekend is Sat 21 – Sun 22 Feb. If you want to build something that matters, meet people you actually want to work with, and pitch to a packed room of health, tech and venture leaders, grab a ticket soon. Spots are limited by design. 👉 Tickets here.

🔥 Exceptional Events (OMG there’s too many)

1. Melbourne | AI Builder Co-working x Stone & Chalk

🗓 Saturday 21st Feb | ⏰ 9:30am | 📍 Stone & Chalk

This event is great for those who want to work on their AI products, see what others are building, get to know Melbourne’s awesome AI community more, and just hang out and have a great time!

👉 Free tickets here

2. MYMI x MLAI: MedHack Frontiers

🗓 21st–22nd February 2026

Australia’s most chaotic health-tech hackathon. Team up with Hackers, Hustlers, Hipsters & Healers to solve real medical challenges and push digital health beyond buzzwords.
👉 Tickets here.

3. Fayl Tales Live: What Almost Broke Me

🗓 27th Feb 2026 | ⏰ 5:30 PM | 📍 Stone & Chalk Melbourne

Most startup events are about how people made it. This one is about how they almost didn’t... This is an intimate live event for founders, operators, and builders who are curious about what really happens behind the highlight reel.

👉 Secure your spot here.

4. buildDay: Build & Deploy a Working AI App in 6 Hours

🗓 28th Feb 2026 | ⏰ 8:30 AM | 📍 Stone & Chalk Melbourne

Build a real AI-powered web app in one day. Deploy it live. Take it home. Never built an app before? Perfect. That’s exactly who this is for.

👉 Secure your spot here.

5. Elastic “Forge the Future” Hackathon

🗓 March 2, 2026 | ⏰ 9:00 AM |📍AWS Office, Sydney

Join developers, data specialists, designers, product managers, and AI builders for a focused, in-person build day at the AWS office. Work alongside Elastic and AWS experts, take part in hands-on workshops, and build real-world solutions using Elastic Search AI, Security, and Observability.

👉 Learn more and secure your spot here.

6. MedHack: Frontiers Pitch Night

🗓 4th March 2026 | ⏰ 6:00 PM | 📍 Monash College City Campus

After one intense weekend of clinicians, engineers, designers, and students speed-running healthcare problems, we’re closing MedHack: Frontiers the only way that makes sense: put the prototypes on stage and make the teams defend them in public (politely, mostly).

👉 Register here.

7. Vibe Marketing: Hack Your Way To Page #1 on Search Engines

🗓 7th March 2026 | ⏰ 10:30 AM | 📍 Stone & Chalk Melbourne

Get your business on page #1 on Seach Engines and ChatGPT while you sleep basically. Come spend a Saturday with us and learn a fast, practical system that works. This workshop is built for founders, business owners & marketers. No fluff. No jargon. Just a clear playbook and code you can use the same day.

👉 Only 12 spots available each month

🤖 AI Bits for Techies

This week: AI Productivity Reality

AI is moving SO FAST! So we moved the heavy lifting to a dedicated weekly report. If you want the deep dive, including weekly research paper summaries, new open-source tools, and model architecture updates, this is for you. This week find out if Moltbook is really a devil or angel.

View the Techies Report: Click Here

📼 Weekly Video: The MoltBot security situation is insane

Length: 11.24mins

About: The takeaway from the video is pretty blunt: we’re currently in the ‘Wild West’ phase where convenience has completely lapped security. If you’re letting an agent browse the web and you haven’t sandboxed it into oblivion, you’re basically leaving your front door open and inviting every weasel on the internet to have a go at your files.

🔮 Coming Next Week: The Distribution Trap

🚀 Topic: Your First 100 Users

We are living in the Golden Age of Shipping. Between AI coding agents, no-code backends, and instant cloud infrastructure, the technical barrier to entry hasn’t just lowered, it has collapsed.

You can now build a SaaS product in a weekend. But can you find a user in a month?

While your dev tools are living in 2026, most go-to-market strategies are stuck in 2010. The result? A marketplace flooded with incredible, high-quality tools that are screaming into the void.

The uncomfortable truth we’re tackling next week: The hardest part of a startup is no longer the stack. It’s the distribution. We are going to show you how to break through the noise when everyone is a builder.

🔔 Subscribe so you don’t miss the learning. Got a growth hack that actually worked? Drop it in #general in Slack (or email us at hi@mlai.au if the link acts up) and collect your Roo Points. 🦘