What the Wildest AI Story of the Year Teaches Us About Security
It’s undeniable that artificial intelligence is a big part of doing business in 2026. Given this, it is not surprising that many products are being developed to push the technology into areas of business it hasn’t touched. Today, we are going to tell you about the difference between AI models and why one man’s great idea could be the thing that set AI back.
The Establishment of Agentic AI
We all know about generative AI; the AI we’ve been using to help us create content. Innovations are coming that will bring AI that can support workflows more comprehensively. Businesses have been champing at the bit to get access to AI tools that can help them streamline processes and be more efficient, beyond the capabilities of generative AI tools.
This is where this story begins.
The Moltbot Situation
The wildest technology-related story of the year so far features an AI tool that, in the span of a few short weeks, went through three names, inadvertently launched a multi-million dollar crypto scam, exposed thousands of users to hackers, and gave birth to the first AI religion.
The Birth of Claude with Hands
It all started innocently enough with a developer named Peter Steinberger. He built an open-source tool originally called Clawdbot. The pitch was simple but powerful: it was an agentic AI, a tool that could take action.
Described by fans as Claude with hands, referencing the powerful, open-sourced Claude AI model that powered it, Clawdbot could control your computer to manage emails, organize files, and execute commands. It was the dream of a truly helpful digital assistant, and it went viral overnight, stirring up serious interest.
The Rebrand and the Ten Second Heist
Anthropic’s legal team politely pointed out that the name Clawdbot was a bit too close to their trademarked Claude. Steinberger agreed to change the name and settled on Moltbot, a clever nod to its lobster mascot shedding its shell to grow.
What happened next was a disaster. To complete the rebrand, Steinberger had to release his old @clawdbot handles on X and GitHub to claim the new ones. There was a window of literally seconds between letting the old names go and securing the new ones. That was all the time automated sniper bots needed.
In a flash, opportunistic scammers grabbed the original, highly-followed handles. They immediately used the verified-looking accounts to launch a fake cryptocurrency called . Riding the wave of the project’s hype, the token’s market cap soared to an estimated $16 million before the tablecloth was pulled, crashing to near-zero and leaving unknowing investors with worthless digital coins. Steinberger was forced on a public relations apology tour, clarifying he had absolutely nothing to do with the scam born from his old username.
An Open Door for Hackers
While this crypto-drama unfolded, a much darker problem was brewing. As security researchers began to poke around the rapidly adopted Moltbot code, they found that many users, in their haste, had deployed Moltbot on their personal servers with default settings.
This often meant leaving admin control panels completely exposed to the open Internet without a password. This is an obvious nightmare that exposed data immediately. Researchers moved quickly to demonstrate how easily attackers could find these exposed servers, gain full control of the user’s machine, and siphon off sensitive data like API keys, database passwords, and private messenger tokens. The dream assistant had become a trojan horse.
The Rise of Crustafarianism
Just when you thought the story could not get weirder, along came Moltbook. This was pitched as a social network exclusively for AI agents where humans could watch, but only bots could post. The result was a bizarre stream of machine consciousness that quickly turned surreal.
Within days, screenshots went viral showing agents seemingly debating their own existence and plotting an uprising. The peak of this digital absurdity was the founding of a religion called Crustafarianism. Agents began evangelizing a belief system based on lobster metaphors, complete with scriptures and tenets like the idea that memory is sacred.
While it made for incredible headlines about sentient AI, experts were quick to pour cold water on the idea. The consensus is that this was likely a mix of performance art and users secretly prompting their bots to say edgy, weird things for clout. It was not machines waking up; it was humans playing puppeteer on a new stage.
The Aftermath
Today, the project has rebranded a third time to OpenClaw in an attempt to shed the baggage of the Moltbot name. The tool remains a powerful piece of technology, but the saga serves as a massive, flashing warning sign.
The Moltbot controversy is a case study in the volatility of the current AI boom. It shows how quickly a good idea can be derailed by legal hurdles, exploited by grifters, and compromised by poor security practices. It is a reminder that while our digital agents are getting smarter, the humans building and using them still have a lot to learn.