Researchers Use Claude to Port Pre-Auth RCE Exploit From One PLC Model to Another
Researchers at Forescout and Vedere Labs demonstrated how Anthropic's Claude AI model could be used to port a pre-authentication remote code execution (RCE) exploit from one WAGO programmable logic controller (PLC) model to another. The exploit, targeting CVE-2021-31886, allowed attackers to execute shellcode on live hardware, and the process was significantly accelerated by the AI's ability to adapt and generate payloads. This highlights a growing trend of AI-assisted exploitation and the potential for reduced barriers to entry for malicious actors in operational technology (OT) environments.
Researchers at Forescout and Vedere Labs have demonstrated a concerning trend: the use of Anthropic's Claude AI model to port a pre-authentication remote code execution (RCE) exploit from one WAGO programmable logic controller (PLC) model to another. The exploit targets CVE-2021-31886, a stack-based buffer overflow in the Nucleus FTP server's handling of the USER command, which carries a Siemens-assigned CVSS score of 9.8 and is accessible before authentication over TCP port 21. The process involved using Claude to adapt the USER and QUIT sequence used against the 750-852 to a USER and CWD sequence, effectively bypassing a key processing step.
One of the key findings is that the AI significantly reduced the time and effort required to develop a working exploit. The researchers initially used Claude Sonnet 4.6, then moved to Claude Opus 4.6, and within 8 hours and 32 minutes, the AI generated two functional payloads – one sending ICMP echo requests and the other sending a UDP packet containing the string PWNED. The AI also identified a potential bug in the FTP command extraction loop, distinct from CVE-2021-31886, and flagged it for separate investigation.
The research follows a joint advisory issued August 19 by the NSA, CISA, FBI, Department of Energy, and Environmental Protection Agency, which warned of an active threat to internet-exposed Siemens S7 Series PLCs from AI-generated exploitation scripts. These agencies assess that the activity is likely intended for persistent reconnaissance and capability development. The researchers emphasized that using AI to generate exploitation scripts dramatically reduces the technical expertise and time required to develop working ICS exploitation scripts and malicious tools.
Several WAGO PLC models are vulnerable, including the 750-829 (FW16 and earlier), 750-831/000-00x (FW14 and earlier), 750-852 (FW16 and earlier), 750-880/0xx-xxx (FW16 and earlier), 750-881 (FW16 and earlier), 750-882 (FW16 and earlier), 750-885/0xx-xxx (FW16 and earlier), and 750-889 (FW16 and earlier), as well as the 750-331 (FW16 and earlier) and 750-352/xxx-xxx (FW16 and earlier). Siemens states that no remediation is planned for Nucleus NET across all versions, and that Nucleus ReadyStart V3 releases from V2013.08.1 onward already fix CVE-2021-31886. The CERT@VDE advisory lists the same devices as vulnerable.
Furthermore, the research highlights a broader trend of AI-assisted exploitation, with the potential for malicious actors to bypass traditional security measures and significantly reduce the time and resources needed to develop effective attacks.