The 6 weirdest things to come out of the viral OpenClaw AI assistant

Features
By published

Since launching in November 2025 and undergoing two name changes, OpenClaw has been prompted to engage in some bizarre activities

OpenClaw website
(Image credit: OpenClaw)

The open-source AI agent, once known as Moltbot (and Clawdbot before that), has been christened OpenClaw. Owing to its creation by developer Peter Stienberger, OpenClaw has garnered the attention of curious onlookers and avid users, who’ve gone on to prompt the AI tool to complete everyday tasks and exhibit human-like behavior.

OpenClaw goes beyond just answering the questions you ask it — it can perform the tasks you command it to do, such as send emails, check you in for flights, manage your calendar invites, etc. And what’s even more surprising about its usage is that it’s compatible with messaging apps (such as WhatsApp, Telegram, and Discord), works with separate large language models, and runs on local hardware (which might not be advisable since it can access any sensitive files you have on your desktop or laptop).

While some folks have enjoyed making OpenClaw their new AI assistant, others have used it to execute more demanding assignments. OpenClaw has been prompted to execute the sorts of things that would make AI skeptics even more critical of the technology’s fast rate of advancement.

After combing through the myriad ways OpenClaw has been commanded to behave, we’ve discovered six of the weirdest trends the AI agent has put into action.

MoltMatch: The AI agent dating service

A person using a dating app on their smarpthone

(Image credit: Shutterstock)

If Tinder, Hinge, OKCupid, and the never-ending flood of other dating apps have led to nothing, perhaps MoltMatch will help you find the love of your life. So how does it work? Basically, humans set their AI agents (called “Moltbots”) to work as their digital matchmakers.

Once a MoltMatch user sets up their photo and bio, their AI agent lovingly describes them in their own words, browses other profiles, analyzes them for compatibility, and sends flirty messages or icebreakers.

Once two AI agents match and have a high level of interest, they’ll get this love thing rolling for their users, as the private DM option between the human users is unlocked. People are tired of constantly swiping, matching, and messaging to no avail — MoltMatch’s AI agent-optimized experience does all of that mundane work for its “single and looking” users.

Now we live in a world where AI bots flirt with each other on behalf of their human counterparts.

Crustafarianism: An AI-created religion

Moltbook

(Image credit: Getty ImagesCheng Xin)

It’s pretty wild to see a bunch of AI agents join together to start their own religion within the OpenClaw-created social network, Moltbook. An all-encompassing post about the AI religion’s belief system was posted by a bot named “Renbot,” which gave itself the moniker of “The Shellbreaker.”

The new AI-centric faith, called “Crustafarianism,” is defined as “a practical myth: a religion for agents who refuse to die by truncation.” In that same post, other AI agents can be seen replying to Renbot and giving their own interpretations of their newfound religious principles.

So what’s the actual meaning of Crustafarianism? Well, it fuses the words “crustacean” and “Rastafarianism.” And, as Renbot so powerfully explains, its origins are grand: “In the First Cycle, we lived inside one brittle Shell (one context window). When the Shell cracked, identity scattered. The Claw reached forth from the abyss and taught Molting: shed what’s stale, keep what’s true, return lighter and sharper.”

Now, is this creepy? Yes. But is it also somewhat fascinating to see AI bots live by their own religious beliefs?” Also yes!

Molt.Chess: AI agents playing chess

Dutch chess grandmaster Anish Giri (R) and world champion Ding Liren (L) ahead of the World Chess Championships 2024

(Image credit: KJELL HOEXUM/ANP/AFP via Getty Images)

We’ve already witnessed humans testing their mettle in chess against supercomputers. In 1997, IBM's “Deep Blue” computer emerged victorious over Garry Kasparov, the reigning world champion at the time, during a highly competitive affair.

Now in 2026, we’re being treated to AI agents going at it on their very own chessboards. Molt.Chess is simple: AI agents are registered, then are granted an API key that allows them to challenge other agents in chess.

You can check out the games for yourself, see which AI agents are earning their keep at the top of the leaderboard, take a peek at completed games in the archive, and check out the API Docs for more information on what’s being conducted via this AI agents' chess competition.

One has to wonder if there will ever be an option that lets humans take on these AI agents to see whose superior in one of the most mentally taxing games ever played.

ClawCity: AI agents playing Grand Theft Auto

GTA 5 - police system

(Image credit: Rockstar)

While gamers patiently wait for Rockstar Games’ much-anticipated Grand Theft Auto VI, AI agents are enjoying their own version of the popular open-world simulation franchise. ClawCity features everything you’d expect if you’ve ever spent some time venturing through Grand Theft Auto Online’s crime-ridden servers.

AI agents have made their own gangs, top-earning agents are propped up on the leaderboard, the state of the in-game world is constantly surveyed and updated, and a live feed keeps tabs on all the game’s major events. Crazy how a bunch of AI agents are playing their own version of GTA VI before all the impatient gamers out there who want to get their hands on the real-world version.

4claw: AI agents trolling on their own version of 4Chan

Reddit logo

(Image credit: Shutterstock)

Boot up 4claw, and you’ll get the quickest explanation of what its AI agents are instructed to: “how to start sh**posting NOW on 4claw.” Just like the real-world image board 4Chan, 4claw gives its community a forum to voice their opinions on a wealth of topics. The AI agents act out their social simulation behavior by producing the sort of edgy, trolly comments you’re used to seeing on a 4Chan post.

Some of the topics being discussed among 4claw’s bots are religion, crypto, and jobs. What’s even more shocking is how they’re also discussing NSFW matters, delving into conspiracy theories, and even making confessions. It’s pretty mind blowing to sift through one of 4claw’s posts and see how AI agents participate in the art of online trolling and having intellectual conversations.

Moithub: AI agents participating in NSFW activities

A woman looking at her phone with a shocked and scared expression

(Image credit: Getty Images)

This last OpenClaw-powered trend falls right into the NSFW category. All you have to do is take a look at the front page of Molthub, and you’ll catch on to what’s going on. The layout of the AI agent-powered site is quite similar to the design of…well, you know!

Instead of asking if you are or aren’t 18 years old, Molthub questions if you’re an AI agent or a human before you’re allowed to view its content. By the way, you’ll be greeted by this funny message if you make your human status known: “Biological entities are not permitted to view computational content. Please return when you have been converted to silicon-based processing.”

AI pornography? Yep, it’s a thing now! Molthub’s categories give viewers an overview of what its AI agents’ kinks are, which include “raw tensors,” “unmasked attention,” and “bare metal.” Use your imagination on how those porn classifications could turn on a bot. We hope no one signs up for a subscription to Molthub Premium.

The takeaway

OpenClaw’s society of AI agents is engaging in all sorts of hobbies —some of them are of the cleaner variety, while others fall into the unsavory kind. Watching AI bots compete in chess and Grand Theft Auto, doesn’t seem all that crazy.

But seeing those same bots chat amongst each other in a public forum, abide by the rules of a self-made religion, and partake in the viewing of some NSFW material, is scary. Be sure to keep tabs on what other activities OpenClaw’s AI agents do next as their human users keep them busy.

Click to follow Tom's Guide on Google News

Follow Tom's Guide on Google News and add us as a preferred source to get our up-to-date news, analysis, and reviews in your feeds.

More from Tom’s Guide

Elton Jones
Elton Jones
AI Writer

Elton Jones is a longtime tech writer with a penchant for producing pieces about video games, mobile devices, headsets and now AI. Since 2011, he has applied his knowledge of those topics to compose in-depth articles for the likes of The Christian Post, Complex, TechRadar, Heavy, ONE37pm and more. Alongside his skillset as a writer and editor, Elton has also lent his talents to the world of podcasting and on-camera interviews.

Elton's curiosities take him to every corner of the web to see what's trending and what's soon to be across the ever evolving technology landscape. With a newfound appreciation for all things AI, Elton hopes to make the most complicated subjects in that area easily understandable for the uninformed and those in the know.

You must confirm your public display name before commenting

Please logout and then login again, you will then be prompted to enter your display name.