OpenClaw is a free and open-source autonomous artificial intelligence agent that can carry out tasks using large language models (LLMs). Its main user interface is a messaging platform, so users interact with it through a chatbot inside services such as Signal, Telegram, Discord, or WhatsApp. The project drew wide attention in early 2026, both for its flexibility and for the security and privacy concerns raised by the broad permissions it requires to operate.
Development and naming
Austrian programmer Peter Steinberger first published the software in November 2025 under the name Warelay. It was derived from an earlier assistant he had built, Clawd, whose name referred to Anthropic's chatbot Claude. The project was renamed twice within about two months, first to Moltbot after trademark complaints from Anthropic, and then to OpenClaw. In February 2026, Steinberger announced he would join OpenAI and that a non-profit OpenClaw Foundation would take over stewardship of the project.
Functionality and security
OpenClaw runs locally and connects to external models such as Claude, DeepSeek, or OpenAI's GPT family, storing configuration and interaction history on the user's machine. A skills system tells the agent how to perform tasks and call tools. Because the software can access email, calendars, and messaging accounts, misconfigured instances and prompt injection attacks have drawn scrutiny from cybersecurity researchers, and some governments have restricted its use in official settings.