OpenClaw AI Runs Wild in Business Environments

OpenClaw, the open-source AI agent previously known as MoltBot and ClawdBot, has become the fastest-growing project on GitHub, with its adoption surging roughly 14-fold in the past week and more than 113,000 stars, up from about 7,800 on 24 January.

The project connects to email, files, messaging platforms and system tools, creating persistent non-human identities and access paths that fall outside traditional IAM and secrets controls, and it can execute terminal commands, read and write files, browse the web and retain memory across sessions, according to Token Security.

Security researchers warn that giving such agents privileged access to local applications and chat channels introduces significant risks, with Pillar Security noting attackers scanning for the default MoltBot/OpenClaw and Ox Security highlighting supply‑chain risks and the potential for a data-breach scenario if a single compromised machine or malicious update exposes access to multiple connected accounts.

The project has about 350 contributors, and some experts warn that a single malicious commit could introduce a backdoor affecting more than 300k users, who have granted OpenClaw access to private platforms like Gmail and WhatsApp. Despite these concerns, developers and researchers such as Dan Guido of Trail of Bits and Ido Shlomo of Token Security emphasise ongoing collaboration and the importance of strong identity controls and secure deployment practices. 30 January 2026.