Researchers have detailed how a solo Russian-speaking threat actor, tracked under the alias bandcampro, integrated Google’s open-source Gemini CLI tool into their offensive workflow to help operate a small botnet built from compromised dental clinic computers.
The findings stem from an analysis of roughly 200 Gemini CLI session logs collected between March 19 and April 21, 2026. According to the analysis, the actor used the AI tool to assist with several stages of the attack chain, including cracking passwords and configuring infrastructure to route traffic through what appears to be a residential proxy setup, likely to mask the origin of malicious activity.
A Small but Notable Botnet
The compromised botnet reportedly consisted of eight machines belonging to dental clinics, a modest footprint compared to large-scale criminal botnets, but notable for illustrating how a single operator can lean on freely available AI tooling to handle tasks that would otherwise require deeper technical expertise or additional manual effort.
AI as a Force Multiplier for Low-Skill Operators
The case underscores a growing trend security researchers have been tracking: threat actors, including those without advanced technical backgrounds, increasingly outsourcing operational tasks to command-line AI assistants. Tools like Gemini CLI, designed to help developers automate coding and system administration tasks through natural language prompts, can just as easily be repurposed to script password attacks, manage proxy configurations, and orchestrate command-and-control activity for a botnet.
Because Gemini CLI is open-source and freely available, there is little friction preventing threat actors from adopting it as an operational aid rather than relying solely on custom-built or purchased malicious tooling.
Why It Matters
While the scale here is small, the technique is significant. It shows real-world evidence of AI command-line tools being used operationally by cybercriminals rather than merely tested or discussed theoretically. Security teams, particularly those defending small organizations like clinics with limited IT resources, should treat weak credential hygiene and exposed remote access services as high-priority risks, since these remain the likely entry points such automated tooling is used to exploit.
Defenders are encouraged to monitor for anomalous outbound connections consistent with residential proxy traffic and to enforce strong, unique credentials and multi-factor authentication on internet-facing systems.