#491 – OpenClaw: The Viral AI Agent that Broke the Internet – Peter Steinberger - Lex Fridman Podcast Recap

Podcast: Lex Fridman Podcast

Published: 2026-02-12

Duration: 0 min

Guests: Peter Steinberger

Summary

In this episode, Peter Steinberger discusses OpenClaw, an open-source AI agent designed to operate autonomously on your computer. This AI assistant represents a significant advancement in the field of AI, transitioning from mere language processing to actionable agency.

What Happened

Lex Fridman sits down with Peter Steinberger, the mind behind OpenClaw, to explore this groundbreaking AI tool that has been dubbed one of the biggest moments in AI since the launch of ChatGPT. Originally known as Moldbot, the name was changed at the request of Anthropic to avoid confusion with their AI model, Claude. OpenClaw is described as an autonomous assistant that can communicate through various messaging platforms and utilize multiple AI models, including Claude Opus 4.6 and GPT 5.3 Codex, to perform tasks for users.

The conversation delves into the potential and risks associated with OpenClaw. While it offers users unprecedented control over their data and the ability to complete tasks autonomously, Peter warns that such power comes with significant security concerns. The agentic capabilities of OpenClaw mark a pivotal shift in how we interact with AI, making it feel more like a personal assistant that understands and learns from its user. Peter reflects on his journey, having spent 13 years developing PSPDF Kit, taking a hiatus from programming, and ultimately returning to create this influential AI agent in a remarkably short time.

As they discuss the implications of this technology, Peter emphasizes that OpenClaw signifies the dawn of the agentic AI revolution. He compares this moment to past milestones in AI development, suggesting that we are now living through the 'age of the lobster.' This episode not only highlights the innovative features of OpenClaw but also prompts listeners to consider the ethical responsibilities that come with such powerful technology.

Key Insights

Key Questions Answered

What is OpenClaw and how does it work?

OpenClaw is an open-source AI agent designed to act autonomously on behalf of its users. It interacts through messaging platforms like Telegram and WhatsApp, utilizing various AI models to perform tasks and assist users. Its unique capability lies in its ability to access user data with permission, which enhances its functionality as a personal assistant.

Why was the name OpenClaw chosen?

The name OpenClaw was adopted after a request from Anthropic to avoid confusion with their AI model named Claude. The playful reference to 'Lobsterclaw' embodies the agent's functionality while distinguishing it from other AI models, marking a critical branding decision in its development.

What are the risks associated with using OpenClaw?

While OpenClaw provides users with enhanced control over their data and tasks, it also presents significant security risks. Steinberger warns that granting an AI agent system-level access can create vulnerabilities that users need to be aware of. The powerful capabilities of OpenClaw come with a responsibility to manage and secure personal information to prevent misuse.

How does Steinberger's background influence OpenClaw?

Peter Steinberger spent 13 years developing PSPDF Kit, which was widely used across numerous devices. After a three-year hiatus from programming, he returned with a renewed passion, leading to the rapid development of OpenClaw. His extensive experience in software development positions him as a key figure in the current AI revolution in programming.

What does the future hold for AI agents like OpenClaw?

Steinberger suggests that we are entering a new era of agentic AI, with OpenClaw marking a significant milestone. He envisions a future where AI agents can seamlessly integrate into our daily lives, enhancing productivity and personal assistance while emphasizing the need for responsible usage and security measures to protect users.