Loading story…
Forescout Vedere Labs used Anthropic’s Claude Code to adapt a known remote-code-execution exploit from a WAGO 750-852 programmable logic controller (PLC) to the related WAGO 750-831. The exploit abuses CVE-2021-31886, a pre-authentication buffer overflow in the Nucleus FTP server that enables unauthenticated execution of arbitrary ARM shellcode. With terminal access, Ghidra, reference files, and a physical PLC, researchers confirmed the flaw and produced working payloads, although human researchers had to resolve false leads and supply reverse-engineering context.
The initial exploit-porting effort took eight hours and 32 minutes and cost $535.74 in API usage, indicating AI remained less efficient than a skilled human expert for this task. After code execution was achieved, however, the model created working network payloads within minutes; a later attempt to create a command-and-control implant bricked the test PLC after writing to flash-mapped memory. Researchers warned that improving models and faster post-breakthrough iteration could lower the expertise and cost needed to adapt embedded-system exploits, raising operational-technology and critical-infrastructure risk.
See affected versions and whether adversaries are exploiting it.
6 events from the most recent confirmed update back to the earliest known activity.
WAGO reported that multiple controller models are affected by NUCLEUS TCP/IP vulnerabilities that could let an attacker with device access manipulate or disrupt equipment. It listed models affected by all referenced flaws and a second group affected only by CVE-2021-31344, CVE-2021-31346, and CVE-2021-31890, while stating that WAGO devices are not affected by CVE-2021-31885.
Siemens published security advisory SSA-044112 addressing multiple vulnerabilities in the TCP/IP stack of Nucleus RTOS (NUCLEUS:13).
In a separate effort to extend the RCE into a command-and-control implant, Claude tested increasingly complex payloads. One payload wrote to flash-mapped memory and permanently bricked the physical WAGO PLC target.
After researcher guidance and changes to preserve injected code that was being erased before execution, Claude generated working payloads for the WAGO 750-831. The RCE-development stage took more than eight hours, including an eight-hour-and-32-minute reported exercise costing $535.74 in API usage.
Researchers found that normal FTP processing cleared the buffer containing injected shellcode after the oversized USER command. They preserved the buffer by following USER with CWD without a CRLF terminator, enabling arbitrary ARM shellcode execution on the WAGO 750-831.
Forescout Vedere Labs used Claude Code, reverse-engineering tools, and a live WAGO 750-831 PLC to verify that CVE-2021-31886 could be adapted from the WAGO 750-852. An initial payload crashed the PLC, confirming the vulnerability but not delivering controlled execution.
Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.
See whether adversaries are exploiting this yet, and where the affected versions run in your environment.
9 references tracked. Mallory keeps watching after this page renders.
cybersecuritynews.com
Open sourcesecurityaffairs.com
Open sourcethehackernews.com
Open sourceitsecurityguru.org
Open sourcesecurityweek.com
Open sourceforescout.com
Open sourceitpro.com
Open sourcecertvde.com
Open sourcecert-portal.siemens.com
Open sourceMap indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
Reports claim Anthropic’s Claude was used to identify and weaponize **CVE-2026-4747**, a FreeBSD kernel vulnerability in the `RPCSEC_GSS` component used for Kerberos authentication in NFS servers. The flaw reportedly allowed an unauthenticated attacker to trigger a stack buffer overflow and gain remote code execution, ultimately obtaining a **root shell** on unpatched systems. A social media post and a Forbes report both describe the incident as a notable milestone in AI-assisted offensive security, with the advisory crediting Nicholas Carlini using Claude for discovering the bug. According to the Forbes account, an autonomous AI workflow analyzed the published advisory and produced two working exploits in roughly four hours, moving from vulnerability identification to practical exploitation without the extended manual effort typically required for kernel exploit development. The report argues that this sharply reduces the time between disclosure and weaponization and raises pressure on defenders to accelerate patching, continuous code auditing, and AI-assisted detection and response as similar systems are reportedly being applied to find hundreds of additional high-severity flaws.
Jun 29, 2026
Unit 42 investigated a ransomware incident in which a human-directed threat actor used frontier AI models and attack-specific agentic frameworks to compromise an enterprise network in under 10 hours—activity assessed to resemble roughly two weeks of coordinated red-team work. After exploiting a public-facing API endpoint, the automated operation used more than 50 MITRE ATT&CK techniques to enumerate internal services, harvest repository secrets, compromise the organization’s secrets-management platform, and abuse CI/CD workflows. The attackers used stolen cloud credentials to access the victim’s AI infrastructure and attempted to establish Terraform backdoors, although branch-protection controls blocked that persistence attempt. The operators also left an 80-page technical audit detailing the weaknesses they had exploited, underscoring how AI agents can rapidly automate established intrusion tradecraft without relying on a zero-day flaw or exceptional operator capability.
Sep 5, 2026
The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added CVE-2021-26829, a cross-site scripting (XSS) vulnerability in OpenPLC ScadaBR, to its Known Exploited Vulnerabilities (KEV) catalog following evidence of active exploitation. This flaw affects OpenPLC ScadaBR through version 1.12.4 on Windows and 0.9.1 on Linux, specifically via the `system_settings.shtm` component. The vulnerability allows attackers to manipulate the HMI login page and system settings, potentially disabling logs and alarms, which could have significant operational impacts on industrial control systems. Recent reports indicate that the pro-Russian hacktivist group TwoNet exploited this vulnerability against a honeypot mimicking a water treatment facility. The attackers gained initial access using default credentials, established persistence by creating a new user account, and then leveraged CVE-2021-26829 to deface the HMI interface and disrupt system monitoring. CISA's alert underscores the ongoing risk to industrial environments and the need for immediate remediation of affected OpenPLC ScadaBR installations.
Jun 29, 2026