Threat researchCisent
A Gentlemen affiliate used an AI coding assistant as a command channel
CloudSEK documented how an attacker gave orders through a coding assistant's tools, in real intrusions against organizations in six countries.

On October 5, 2026, the threat intelligence company CloudSEK published its research on “Azazel”, a Russian-speaking affiliate of the ransomware group The Gentlemen. In real intrusions, the affiliate used the Model Context Protocol (MCP), which AI assistants use to call tools, to run commands: the tool interface of an AI coding assistant became its command-and-control channel.12
What happened
CloudSEK found the activity while investigating the attacker’s exposed infrastructure: roughly 50 TB of data, with folders for more than two dozen victims in six countries.2
Most of the intrusions began with the theft of credentials stored in GitLab: CI/CD variables and repository history. Another began on an AI platform, through an endpoint that fetched web addresses from the server without validating them properly.2
The victims are in logistics, insurance, pharmaceuticals, artificial intelligence, medical devices and government-adjacent infrastructure. According to CloudSEK, Azazel also ran a leak site of its own, LEAKNED, which it allegedly used to divert extortion payments owed to The Gentlemen’s operators.2
Why it matters
AI assistants already work inside systems, with permission to run tools. If an attacker reaches that interface, its orders travel through a channel that looks like normal work. CloudSEK recommends treating MCP execution tools as privileged interfaces, auditing their use and limiting who can invoke them; not exposing services beyond the machine itself; and rotating GitLab secrets across repository history too.2
Where Cisent comes in
Sentinel Red Team attacks the organization’s systems with its authorization and within the scope it defines, to find the gaps before an attacker does: the secret forgotten in a repository’s history, the service exposed beyond what it needs, the AI tool with more permissions than it requires. It’s ongoing testing, not one audit a year, and every new attack pattern helps train the defense.
Sources
CloudSEK, “Caught in 4K: The Gentlemen Files”, October 5, 2026. https://www.cloudsek.com/blog/caught-in-4k-the-gentlemen-files↩︎
RH-ISAC, “The Gentlemen Ransomware Affiliate Utilizes MCP in Multi-Sector Cyberattacks”, October 6, 2026. https://rhisac.org/threat-intelligence/the-gentlemen-ransomware-affiliate-utilizes-mcp-in-multi-sector-cyberattacks/↩︎