Now on its third name, this agent can take actions without you having to prompt it and make those decisions by accessing large swaths of your digital life.<\/b>
\nUPDATE 1\/30: Third time’s the charm? Moltbot (previously Clawdbot) got yet another name change overnight and is now known as OpenClaw.
Moltbot was „chosen in a chaotic 5am Discord brainstorm with the community“, says creator Pete Steinberger, „but it never quite rolled off the tongue.“ So, it’s going with OpenClaw. „And this time, we did our homework: trademark searches came back clear, domains have been purchased, migration code has been written. The name captures what this project has become.“
Original Story 1\/28:
AI agents are hit or miss, but a lobster-inspired assistant has piqued the interest of developers and vibe coders alike to become the internet’s latest AI obsession. Initially known as Clawdbot, it’s now Moltbot following a complaint from Anthropic over similarities in name to its Claude AI.
It’s not particularly common for an open-source AI tool to go viral, given its fairly niche audience and the technical know-how required to set it up on GitHub. So, this one caught our attention.
Moltbot is free to download, but it’ll cost about $3\u2013$5 per month to run on a basic Virtual Private Server (VPS). Some people have had success setting it up on AWS’s free tier. Contrary to the impression social media posts can give, you do not need an Apple Mac mini to run it, according to creator Pete Steinberger. Moltbot will run on any computer, including that old laptop collecting dust in your closet.
Steinberger’s X bio claims he „came back from retirement to mess with AI and help a lobster take over the world.“ Yet momentum around agentic assistants largely petered out late last year. Perplexity’s Comet browser felt half-baked and not entirely useful, our analyst Ruben Cirelli found. OpenAI warned that its Atlas AI browser may purchase the wrong product on your behalf, and is vulnerable to prompt injection attacks. Will Steinberger’s tool revive interest? Should it?
The defining features of Moltbot are that it can (1) proactively take actions without you needing to prompt it, and (2) make those decisions by accessing large swaths of your digital life, including your external accounts and all the files on your computer, sort of like Claude Cowork. It might clear out your inbox, send a morning news briefing, or check in for your flight. When it’s done, it’ll message you through your app of choice, such as WhatsApp, iMessage, or Discord.
This open access has raised security concerns. Support documentation even acknowledges that „Running an AI agent with shell access on your machine is\u2026 spicy. There is no ‚perfectly secure‘ setup.“ You can run it on the AI model of your choice, either locally or in the cloud.
„For an agent to be useful, it must read private messages, store credentials, execute commands, and maintain persistent state“, says threat intelligence platform SOCRadar. „Each requirement undermines assumptions that traditional security models rely on.“
SOCRadar recommends treating Moltbot as „privileged infrastructure“ and implementing additional security precautions. „The butler can manage your entire house. Just make sure the front door is locked.“
Some argue that keeping data local enhances security, but Infostealers notes that hackers are finding ways to tap into local data, a treasure trove for nefarious actors. „The rise of ‚Local-First‘ AI agents has introduced a new, highly lucrative attack surface for cybercriminals“, it says. „[Moltbot].offers privacy from big tech, [but] it creates a ‚honey pot‘ for commodity malware.“
The important thing is to make sure you limit „who can talk to your bot, where the bot is allowed to act, [and] what the bot can touch“ on your device, the bot’s support documentation says. Developers have begun sharing steps they’ve taken to shore up security. „Start with the smallest access that still works, then widen it as you gain confidence“, Moltbot recommends.<\/p>\n