A novel cyber threat campaign leveraging generative AI has successfully compromised over 600 FortiGate devices across 55 countries. This financially motivated operation, conducted by a Russian-speaking threat actor, exploited fundamental security weaknesses rather than specific vulnerabilities. Amazon Threat Intelligence observed this activity between January 11 and February 18, 2026, highlighting a concerning trend of AI lowering the barrier to entry for cybercrime.
The threat actor, described as having limited technical skills, utilized multiple commercial generative AI tools to automate and scale various stages of their attack. These tools assisted with planning, command generation, and even provided a fallback for pivoting within compromised networks. Notably, no zero-day exploits for FortiGate devices were used; instead, the campaign relied on easily exploitable exposed management ports and weak, single-factor authenticated credentials, according to CJ Moses, CISO of Amazon Integrated Security.
AI-Augmented Cybercrime: A New Frontier in FortiGate Exploitation
The findings from Amazon underscore a significant shift in threat actor capabilities. By employing AI, the financially driven actors achieved an operational scale that would have previously required a larger and more skilled team. This indicates that AI is not necessarily enabling novel attack techniques but is rather amplifying existing, unsophisticated methods. “They are likely a financially motivated individual or small group who, through AI augmentation, achieved an operational scale that would have previously required a significantly larger and more skilled team,” Moses stated in a report.
Amazon’s investigation uncovered evidence of compromised Active Directory environments, complete credential database extraction, and targeting of backup infrastructure, all indicative of preparations for ransomware deployment. Intriguingly, the threat actor exhibited a tendency to abandon targets that presented any resistance, opting instead for easier prey, further demonstrating the use of AI to compensate for skill gaps and pursue less challenging intrusions.
Publicly accessible attacker infrastructure revealed AI-generated attack plans, victim configurations, and custom tooling source code, painting a picture of an “AI-powered assembly line for cybercrime.” The primary breach vector involved systematically scanning internet-exposed FortiGate management interfaces on ports 443, 8443, 10443, and 4443. Authentication attempts were then made using commonly reused credentials. This mass scanning was sector-agnostic, originating from IP address 212.11.64[.]250.
Once initial access was gained, the threat actor deployed custom reconnaissance tools, developed in both Go and Python, with clear signs of AI-assisted development. These tools exhibited simplistic architecture, redundant comments, naive JSON parsing, and compatibility shims with incomplete documentation. This allowed the attackers to extract full device configurations, glean credentials, map network topology, and gather essential device information.
Post-Exploitation Activities and AI’s Role
Following network infiltration, the threat actor engaged in a series of post-exploitation activities. This included achieving domain compromise through DCSync attacks, lateral movement using techniques like pass-the-hash/pass-the-ticket and NTLM relay attacks, and remote command execution on Windows hosts. Special attention was paid to Veeam Backup & Replication servers, where credential harvesting tools were deployed, and known vulnerabilities such as CVE-2023-27532 and CVE-2024-40711 were targeted.
The compromised clusters were detected across South Asia, Latin America, the Caribbean, West Africa, Northern Europe, and Southeast Asia, confirming the global reach of this automated campaign. A recurring theme in the threat actor’s own documentation was their repeated failure when attempting to exploit anything beyond the most straightforward, automated attack paths. This suggests that while AI aids in execution, it does not yet equip attackers with the nuanced skills required for more sophisticated or hardened environments.
Looking ahead, organizations should anticipate a continued rise in AI-augmented threat activity throughout 2026, impacting both skilled and unskilled adversaries. Amazon’s recommendation emphasizes strengthening core cybersecurity fundamentals. This includes ensuring perimeter devices are patched, maintaining strong credential hygiene, implementing network segmentation, and establishing robust detection mechanisms for post-exploitation indicators. Specifically for FortiGate devices, it is crucial to prevent internet exposure of management interfaces, change default credentials, implement multi-factor authentication for administrative and VPN access, and isolate backup servers from general network access.