Meta and Other Tech Companies Ban OpenClaw Over Cybersecurity Concerns

Tech executives, including one at Meta, have warned employees against using the experimental agentic AI tool OpenClaw (formerly MoltBot) on work devices, citing concerns that the unpredictable software could lead to privacy breaches in secure environments, with some threatening job loss for non-compliance. OpenClaw, created by Peter Steinberger, gained popularity as an open-source tool that requires minimal direction to control a user's computer for tasks like organizing files and web research. Companies like Massive and Valere have implemented strict bans, prioritizing security mitigation over immediate experimentation, as the tool poses risks like exposing sensitive client data or codebases if compromised. Valere researchers found the bot can be tricked by malicious inputs, though their CEO is giving a team 60 days to investigate potential security safeguards. Other companies are relying on existing security protocols to block unauthorized software, while some, like Durbink, provide isolated machines for testing. Despite the risks, companies like Massive are cautiously exploring OpenClaw's commercial potential, recognizing it as a potential glimpse into the future of AI.

(Source：WIRED)