Meta and other tech companies are restricting use of OpenClaw, an experimental AI agent that can take control of users' computers. The tool, created by solo founder Peter Steinberger, gained popularity after being released as open source last November but has raised serious security concerns among executives who fear it could lead to privacy breaches.

Several companies have implemented bans or strict controls on OpenClaw usage. A Meta executive warned employees they could lose their jobs for using it on work laptops, while Massive CEO Jason Grad issued an internal warning before any employees had installed the software. Valere, which works with organizations like Johns Hopkins University, initially banned the tool entirely before allowing limited research use on isolated machines.

Security researchers have identified vulnerabilities including the potential for hackers to trick OpenClaw into sharing sensitive files. The tool requires basic software engineering knowledge to set up and can interact with other apps to perform tasks like organizing files and conducting web research. While some companies are choosing to rely on existing security measures rather than ban OpenClaw outright, others are moving quickly to mitigate potential risks before investigating the technology further.