UAT-10147: Chinese-speaking adversary integrates agentic AI into post-compromise operations

UAT-10147 is a Chinese-speaking cybercrime group identified by Cisco Talos in 2026 targeting vulnerable Windows and Linux web servers worldwide. The group uses publicly disclosed vulnerabilities for initial access and employs AI-generated operational playbooks, exploit automation scripts, and adaptive troubleshooting to automate post-compromise operations. The actor uses open-source frameworks such as Metasploit, ysoserial, PentestGPT, and DeepAudit, along with privilege escalation exploits like EfsPotato, to deploy malware including BadIIS, QuasarRAT, Gh0stCringe, and SPECTRE implants. On Windows, multi-stage batch scripts download tools, modify Windows Defender exclusions, escalate privileges, and create persistent backdoors and rogue administrator accounts. On Linux, the attacker deploys web shells for persistent command execution. The campaign targets high-value servers in multiple sectors and countries, with a large target list of approximately 170,000 URLs. Talos notes the actor's use of agentic AI to semi-autonomously orchestrate offensive tradecraft at scale, representing an evolution beyond simple AI-assisted scripting.

The campaign enables large-scale compromise of internet-exposed Windows and Linux web servers across multiple sectors and countries. Successful exploitation leads to remote code execution, malware deployment for SEO fraud and data theft, persistent backdoors, privilege escalation, and creation of rogue administrative accounts. The use of AI-driven automation increases the efficiency and scale of attacks, reducing the expertise required to conduct complex intrusions. This results in significant operational risk to affected organizations, including data loss, unauthorized access, and potential disruption of services.

No specific patches are referenced for this campaign; remediation depends on addressing the publicly disclosed vulnerabilities exploited for initial access. Organizations should apply all relevant security updates for their web servers and underlying software promptly. Monitoring for indicators of compromise such as unusual scheduled tasks, unexpected user accounts, and network connections to known malicious IPs is advised. Since the campaign uses known privilege escalation tools and malware, endpoint detection and response solutions should be tuned to detect these behaviors. Talos has not indicated that the threat is mitigated or no action is required; therefore, active defense and patching are recommended.