CYBERSECURITY researchers have uncovered an AI-driven intrusion ecosystem operated by affiliates of the INC and The Gentlemen ransomware groups. Central to the operation is an Autonomous Red Team Kill Chain Framework called Phantom, used in live operations via OpenCode, with a human operator reviewing every step yet tasks executed by AI agents.
The threat actor is described as Russian-speaking and functions as both intruder and data-extortion operator, as well as an infrastructure provider for partners across the ransomware ecosystem. Edge devices and remote-access systems were targeted at scale, including FortiClient EMS, SimpleHelp, cPanel and VPN gateways, with AI-managed knowledge bases tracking victims, stolen credentials and campaign progress.
Initial access is said to rely on exploiting public-facing edge devices, followed by lateral movement, persistence, data exfiltration and eventual ransomware deployment.
SOCRadar notes multiple exploited flaws, notably CVE-2024-57727 (SimpleHelp path traversal) and CVE-2026-35616 (Fortinet FortiClient authentication bypass), CVE-2026-41940 (cPanel/WHM authentication bypass), plus staged or developed exploits such as CVE-2026-15409 (SonicWall SMA 1000 SSRF), CVE-2026-63030 (WordPress wp2shell), CVE-2026-42208 (LiteLLM SQL injection), and CVE-2026-9082 (Drupal SQL injection).
The attackers reportedly reuse credentials widely—one case used a single reused password to unlock VPN, servers, RMM, backup and security platforms. Defensive meaures were evaded with BYOVD components and loader-level evasion, though some enterprise EDRs recorded mixed results.
The operation is supported by a toolkit including Phantom, WormGPT-MCP, Ghost C2, RevSocks, a Mythic/QuickLine-like agent, and exfiltration tools such as rks3 and BackupAgent. Dozens of AI-authored extortion records show a data-centric approach: prioritising financials, then sending masked samples and conducting scripted vishing to pressure payment.
About 62 organisations were targeted, with the U.S. accounting for more than half of victims; sectors hit include industrial, manufacturing and energy, followed by healthcare. The final data copies often appear on data-leak sites if victims do not pay within 24–72 hours, though payment outcomes remain unverified.