A lobster-themed open-source AI bot is going viral for its ability to perform personal tasks, such as managing calendars, sending messages, and even checking flights.
The Moltbot, developed by Austrian developer Peter Steinberger, features the tagline “AI that actually does things” and has amassed over 95,000 GitHub stars. Since several people used Cloudflare’s infrastructure to host the new AI assistant, the cybersecurity company’s shares rose 14%.
Steinberger previously named it Clawdbot since he is a fan of Anthropic’s Claude model.
However, the AI firm sued him under copyright law, and he changed the name to Moltbot while retaining the lobster imagery.
While the AI assistant is convenient, cybersecurity experts warn it has risks.
Because Moltbot executes commands directly on a user’s device, it opens the door to security vulnerabilities like prompt injection. A malicious message sent to a user could trick the AI into running unauthorized commands.
Early adopters are advised to run the agent on separate private servers rather than on personal devices that contain sensitive data.