The rise of AI-powered hacking tools has become a pressing concern for cybersecurity experts. A recent development, the CyberStrikeAI platform, has been identified as a potential game-changer in the hands of malicious actors.
This open-source AI security testing platform has been linked to a hacking campaign that successfully breached hundreds of Fortinet FortiGate firewalls. The campaign, reported by BleepingComputer, compromised over 500 devices in just five weeks, an alarming rate of success.
The threat actor behind this operation utilized multiple servers, including one at 212.11.64[.]250, which was later found to be running CyberStrikeAI. Team Cymru's Senior Threat Intel Advisor, Will Thomas, delved into this discovery, analyzing NetFlow data to uncover the connection.
"CyberStrikeAI's service banner was observed on port 8080, and network communications between this IP and the targeted FortiGate devices were evident," Thomas explained. The infrastructure used in the FortiGate campaign was last seen running CyberStrikeAI on January 30, 2026.
CyberStrikeAI's GitHub repository describes it as an innovative security testing platform, integrating over 100 security tools, an intelligent orchestration engine, and a skills system. It promises end-to-end automation, from vulnerability discovery to attack-chain analysis, all powered by AI agents and a native MCP protocol.
The tool's capabilities are extensive, allowing it to conduct a full attack chain. It can perform network scanning, web and application testing, exploit frameworks, password cracking, and post-exploitation activities. By combining these tools with AI, CyberStrikeAI enables even low-skilled operators to automate attacks.
Team Cymru warns that such AI-native orchestration engines could accelerate the automated targeting of exposed edge devices, including firewalls and VPN appliances. They observed 21 unique IP addresses running CyberStrikeAI between January 20 and February 26, 2026, with servers primarily hosted in China, Singapore, and Hong Kong, and additional infrastructure in the US, Japan, and Europe.
"With adversaries increasingly adopting AI-native orchestration, we can expect a surge in automated, AI-driven attacks on vulnerable edge devices," Thomas said. "Defenders must prepare for a future where tools like CyberStrikeAI, alongside other AI-assisted projects, significantly reduce the entry barrier for complex network exploitation."
The researchers also examined the profile of the CyberStrikeAI developer, known as "Ed1s0nZ." Public repositories linked to the account reveal additional AI-assisted security tools developed by Ed1s0nZ, including PrivHunterAI and InfiltrateX.
Team Cymru's investigation found interactions between Ed1s0nZ and organizations previously linked to Chinese government-affiliated cyber operations. In December 2025, Ed1s0nZ shared CyberStrikeAI with Knownsec 404's "Starlink Project," a Chinese cybersecurity firm with alleged government ties.
The developer's GitHub profile also mentioned receiving a "CNNVD 2024 Vulnerability Reward Program - Level 2 Contribution Award," which is believed to be operated by China's intelligence community. This reference was later removed from the profile.
Ed1s0nZ's GitHub repositories are primarily in Chinese, suggesting a Chinese-speaking developer. While interactions with domestic cybersecurity organizations are not unusual, the connections raise concerns about the potential use of these tools in state-sponsored cyber operations.
The development of these AI-powered cybersecurity tools highlights the increasing use of commercial AI services by threat actors to automate their attacks. Google has also reported that threat actors are abusing Gemini AI across all stages of cyberattacks, empowering hackers of all skill levels.
As malware continues to evolve and become more sophisticated, the need for robust defenses is more critical than ever. The Red Report 2026 reveals how new threats use advanced techniques to evade detection, emphasizing the importance of staying vigilant and adapting security measures.
