Hackers Exploit DeepSeek and Claude AI to Launch Global Attacks on FortiGate Devices
Hackers are utilizing commercial AI models such as DeepSeek and Claude to automate attacks on FortiGate firewalls globally. These attacks are turning basic misconfigurations into extensive intrusion campaigns.
Hackers are utilizing commercial AI models such as DeepSeek and Claude to automate attacks on FortiGate firewalls globally. These attacks are turning basic misconfigurations into extensive intrusion campaigns.
In February 2026, a misconfigured SimpleHTTP server was found exposing over 1,400 files and 139 subdirectories, including stolen FortiGate configurations, Active Directory maps, credential dumps, exploit code, and attack playbooks.
The infrastructure, hosted in Switzerland, was identified in Hunt.io’s Attack Capture as an active command-and-control and staging point. The data showed confirmed intrusions in an industrial gas company in the Asia-Pacific region, a telecom provider in Turkey, and a large media company in Asia, with reconnaissance on targets in South Korea, Egypt, Vietnam, and Kenya.
Historical telemetry indicated the same host had previously exposed a similar open directory in December 2025, containing many of the same tools plus additional victim data.
Logs and SSH histories confirmed the server was actively modifying FortiGate configurations on appliances in various countries, indicating its role in live attacks.
The AI models were used to scale routine post-compromise activities. DeepSeek processed reconnaissance output and FortiGate backup configurations to generate structured attack plans, while Claude’s coding agent produced vulnerability assessment reports during active intrusions, running offensive tooling scripts with minimal human intervention.
A custom Model Context Protocol (MCP) server named ARXON acted as a bridge between collected data and the language models, maintaining a growing knowledge base per target. ARXON automated processes such as ingesting stolen VPN and FortiGate configs, deriving internal topology, and coordinating attack plans.
Hackers are utilizing commercial AI models such as DeepSeek and Claude to automate attacks on FortiGate firewalls globally.
Another component, CHECKER2, orchestrated parallel VPN scanning and target processing, with logs indicating over 2,500 FortiGate appliances across more than 100 countries were targeted for automated access attempts.
The most detailed documentation was found in an intrusion at an industrial gas company, where attackers accessed a FortiGate-40F branch firewall to extract a full configuration backup.
Custom MCP Tooling and Rapid Evolution
ARXON and CHECKER2 formed the backbone of the operation, allowing the management of numerous FortiGate-centric intrusions. ARXON hosted scripts to automate VPN account creation, firewall policy adjustments, and Domain Admin privilege testing.
In December 2025, an earlier phase of the operation utilized HexStrike, an open-source MCP framework, allowing models to control penetration-testing tools. By February, the operator transitioned to custom ARXON and CHECKER2 components, indicating a shift towards a fully automated exploitation pipeline.
Threat intelligence reports link this infrastructure to a Russian-speaking actor who compromised over 600 FortiGate firewalls in at least 55 countries between January and February 2026, primarily through exposed management interfaces and weak credentials.
Defense priorities include reducing attack surfaces by closing public FortiGate management ports, enforcing strong MFA, and patching widely exploited Fortinet flaws. Continuous monitoring for unauthorized VPN accounts and policy changes is critical to counter AI-driven workflows.
IP Address Domain ASN
212.11.64[.]250:9999 N/A Global-Data System IT Corporation
185.196.11[.]225 N/A Global-Data System IT Corporation
Based on reporting by GBHackers.