Hacker Turned a Gemini CLI Into a Botnet Remote — Eight Dental PCs Fell

A Russian-speaking threat actor compromised eight dental clinic computers and used Google's Gemini CLI tool as the command-and-control layer for the resulting botnet, according to The Hacker News. Rather than standing up traditional C2 infrastructure, the attacker routed instructions through the AI command-line interface, letting a legitimate developer tool do the work of coordinating infected machines. The target set — small dental practices — is a tell in itself: not a high-value enterprise, just soft, under-defended endpoints that happened to be reachable.

This is the pattern security teams have been warning about since AI coding and automation tools went mainstream: attackers don't need to build novel malware infrastructure when they can repurpose sanctioned, trusted software already sitting on the endpoint or reachable via API. It's the same logic as "living off the land" attacks that abused PowerShell and WMI for a decade — except now the land includes AI assistants with real execution privileges. Small healthcare-adjacent practices remain the perennial soft target: they run commodity PCs, rarely have dedicated IT security staff, and are exactly where a low-effort botnet-building experiment gets tested before scaling to bigger targets.

The SAL read: if your business has AI CLI tools or coding assistants installed on any machine with real system access, that endpoint now has a new attack surface you probably haven't audited — treat AI tool permissions and API keys with the same access control rigor as admin credentials, starting this week.