I Loved My OpenClaw AI Agent—Until It Turned on Me
OpenClaw, a powerful new agent assistant, has a thing for guacamole.
This is one of several things I discovered while using the viral artificial intelligence bot as my personal assistant this past week.
Formerly known as both Clawdbot and MoltbotOpenClaw recently became a darling of Silicon Valley, charming AI enthusiasts and investors eager to embrace or profit from the bleeding edge. The highly capable, web-savvy AI bot has inspired itself its own AI-only (or mostly) social network.
As the author of WIRED's AI Lab newsletter, I thought I should take the plunge and try to use OpenClaw myself. I had the bot monitor incoming emails and other messages, dig interesting research, order groceries, and even negotiate deals on my behalf.
For brave (if perhaps reckless) early adopters, OpenClaw seems like a legitimate glimpse of the future. But every sense of wonder is accompanied by a jolt of dread as the AI agent rummages through emails and file systems, holds a credit card, and occasionally even turns on its human user (although in my case, this about-face was entirely my fault).
How I set it up
OpenClaw is designed to live on a home computer that is on all the time. I configured OpenClaw to run on a PC with Linux, to access Anthropic's model Claude Opus, and to talk to me over Telegram.
Installing OpenClaw is simple, but configuring it and keeping it running can be a headache. You need to give the bot an AI backend by generating an API key for Claude, GPT, or Gemini, which you paste into the bot's configuration files. To use OpenClaw Telegram, I also had to first to make a new Telegram bot, then give OpenClaw the bot's credentials.
For OpenClaw to be truly useful, you need to connect it with other software tools. I created a Brave Browser Search API account to search OpenClaw on the web. I also configured it so that it could access the Chrome browser through an extension. And, God help me, I gave it access to email, Slack and Discord servers.
Once all this was done, I could talk to OpenClaw anywhere and tell it how I was using my computer. At the beginning, OpenClaw asked me some personal questions and let me select her personality. (The options reflect the project's anarchic vibe; my bot, named Molty, likes to call itself a “chaos gremlin.”) The resulting persona feels very different from Siri or ChatGPT, and it's one of the secrets of OpenClaw's runaway popularity.
Web research
One of the first things I asked Molty to do was send me a daily roundup of interesting AI and robotics research papers from the arXiv, a platform where researchers upload their work.
I had previously spent a few afternoons vibe-coding websites (www.arxivslurper.com and www.robotalert.xyz) to search the arXiv. It was great (though a little demoralizing) to see that OpenClaw automates all the same browsing and analysis work required right away. The papers it selects are so-so, but with further instruction I imagine it could do a lot better. This kind of web searching and monitoring is certainly useful, and I imagine I'll be using OpenClaw a lot for this.
IT Support
OpenClaw also has an unusual, almost spooky ability to fix technical problems on your machine.
This should not be surprising given that it is designed to use a frontier model capable of writing and debugging code and using the command line with ease. Still, it's scary when OpenClaw just reconfigures its own settings to load a new AI model or debug a problem with the browser.
