A Russian-speaking hacker recently used Google’s Gemini CLI, a free, open-source AI coding tool, to build, migrate, and operate a small botnet. Trend Micro documented the whole thing. It’s one of the first real-world cases where a criminal used an AI agent not just to write malware, but to actively run day-to-day operations.

The target? Eight computers in a dental clinic.

What actually happened

The threat actor, tracked as “bandcampro,” ran over 200 sessions with Gemini CLI between May and June 2026. The AI handled the full lifecycle:

Migration: The attacker typed “Study the C2 migration” and the AI read a playbook, prepared the code, deployed it to a VPS, configured Cloudflare tunnels, and debugged connectivity issues. Total time: six minutes.
Daily operations: The attacker asked which machines were online, listed files on specific computers, and generated new infection links. All in plain English.
Troubleshooting: When bots failed to reconnect after the migration, the AI diagnosed conflicting traffic between old and new servers and told the attacker what to do.
Improvements: The AI proposed operational improvements 59 times without being asked.

The entire botnet infrastructure (infection code, persistence mechanisms, and C2 playbook) fit in three plain-text files totaling about 5 KB.

Why a dental clinic matters to your firm

The malware wasn’t sophisticated. No obfuscation, no advanced evasion techniques. Trend Micro described it as basic. The attacker gained access to the OpenDental database, which contains patient records, insurance information, and billing data.

Sound familiar? Swap “dental records” for “client files” and “OpenDental” for your practice management system, and the attack looks the same.

Small professional services offices are targets because they hold concentrated sensitive data and often lack the security infrastructure of larger organizations. A dental clinic. A CPA firm. A 30-person law firm. From an attacker’s perspective, they look identical: valuable data, limited security staff, systems that are often managed reactively rather than proactively.

What changed

This isn’t about a sophisticated attacker doing something clever. That’s what makes it worth paying attention to.

The attacker didn’t need a team or custom tools. They needed a free AI tool and a 5 KB playbook. When something broke, they didn’t debug it themselves. They asked the AI to figure it out.

We wrote a few weeks ago about how AI tools connected to your systems create new doors for attackers to walk through. This is the other side of that same problem: AI tools are also making attackers faster and more capable.

The barrier to running a botnet used to be knowledge and time. That barrier is a lot lower now.

What this means for your IT security

The practical implication isn’t that you need to defend against AI specifically. The attacker’s malware was basic. Standard endpoint protection would catch it. The problem is volume. When the tools to attack are this accessible, more people will use them, and they’ll move faster.

The defenses that matter haven’t changed:

“Are our endpoints actually monitored, or just installed?” The dental clinic’s systems were compromised and polled a command server every five seconds. That’s traffic that shows up in logs, if anyone is looking. Endpoint protection that isn’t actively monitored is just software taking up disk space.

“How quickly would we know if something was wrong?” The attacker operated across 200+ sessions over weeks. Detection speed is what separates “we caught it early” from “they had access for a month.”

“Are our systems patched and configured, or just deployed?” The botnet used scheduled tasks, WMI events, and registry modifications for persistence, all techniques that proper system hardening and configuration monitoring can prevent or detect.

“Do we have someone watching for unusual behavior?” The bots in this case were polling a server every five seconds. That’s an anomaly. But anomalies only matter if someone is actively looking for them.

The takeaway

AI didn’t invent cybercrime. It didn’t create new attack techniques. What it did is make existing techniques available to more people with less effort. A single person with a free tool ran an operation that would have previously required a small team.

Your firm’s security posture doesn’t need to be revolutionary. It needs to be consistently maintained, actively monitored, and regularly reviewed. The basics have always mattered. They just matter more now, because there are more people capable of testing them.

Artech Solutions provides managed security for law firms and professional services firms in the Des Moines metro. If you’re not sure whether your systems are being actively monitored or just passively protected, that’s a conversation worth having.