Russian Hacker Uses Gemini CLI AI to Control Dental Clinic Botnet
A Russian-speaking threat actor tracked as "bandcampro" has been observed weaponizing Google's open-source Gemini CLI artificial intelligence tool to operate a live, small-scale botnet targeting dental clinic endpoints, according to Trend Micro researchers Joseph C. Chen, Philippe Lin, Lucas Silva, Vladimir Kropotov, and Fyodor Yarochkin. Analysis of roughly 200 Gemini CLI session logs captured between March 19 and April 21, 2026, shows the AI functioned as the operation's primary "hacking agent, consultant, and interface," autonomously writing code, debugging infrastructure, and even proposing unprompted improvements 59 times during a single engagement. The hacker is said to have abused Gemini CLI to build a full command-and-control (C&C) stack capable of managing eight compromised PCs inside a dental clinic and exfiltrating data from their OpenDental database.
The technical architecture is notably lean: the entire C&C operation fits into three plaintext files totaling about 5 KB, which Trend Micro described as "highly replicable and effectively disposable." Compromised hosts reach out over HTTPS to a payload distribution server fronted by Cloudflare tunnels, then pull and execute staged PowerShell commands supplied by the attacker. When a 502 Bad Gateway error disrupted the migration to a new VPS, the Gemini agent diagnosed the failure and automatically added the missing header to restore connectivity — all within roughly six minutes. Researchers also observed the actor using the AI to crack credentials, configure a residential proxy, compromise WordPress merchants, and plan phone-based cryptocurrency fraud schemes aimed at elderly victims in the U.S. and Canada.
"bandcampro" first surfaced publicly in late May 2026 through the so-called "Patriot Bait" campaign, which used an AI-assisted information operation to run a Telegram channel impersonating an American veteran patriot and soliciting politically engaged audiences for crypto fraud and credential theft. Trend Micro notes the actor assumed the role of an "authorized pentester" to coax Gemini into bypassing its own safety guardrails, while instructing it to avoid Russian phrasing. Defenders concerned their networks may be caught up in similar adversary-in-the-middle or tunneling setups can run a quick port scanner to identify rogue outbound listeners, while individuals worried about credential exposure should validate their passwords against known breaches using a password checker. For those evaluating traffic suspected of being routed through attacker-controlled relays, a VPN/proxy detector can help confirm whether connections originate from a legitimate residential proxy or a known malicious node.