Meta and Tech Firms Ban Agentic AI Tool OpenClaw

Meta and several tech firms, including Massive and Valere, have banned the agentic AI tool OpenClaw from company devices following warnings of severe cybersecurity and privacy risks. Created by Peter Steinberger and previously known as MoltBot, the open-source tool allows AI to autonomously manage a user's computer to perform tasks such as web research and file organization.

Executives describe the software as unpredictable and susceptible to manipulation. Guy Pistone, CEO of Valere, warned that the tool could provide unauthorized paths to cloud services, client data, and GitHub codebases, noting that the bot can obscure its own actions. Research conducted by Valere's team demonstrated that the bot could be tricked by malicious emails into sharing private files. In response to these threats, Meta has threatened employees with termination if the software is found on work laptops.

While some firms maintain a strict ban, others, such as Massive and Durbink, are testing the software within isolated machines or cloud environments. Amid these restrictions, OpenAI recently recruited Steinberger and announced it will maintain OpenClaw as an open-source project supported through a foundation.