Experiment: Porting a PLC Exploit With AI Takes Hours and Hundreds of Dollars
Forescout researchers conducted an experiment using Claude AI to port a remote code execution exploit across different WAGO PLC models, documenting the time and financial costs involved in the process. The research demonstrates both the capabilities and limitations of leveraging large language models for adapting existing exploits to new targets in industrial control systems.
CISA argues that the security industry's traditional approach of treating vulnerabilities as individual fixes is ineffective against attackers, and advocates instead for eliminating entire vulnerability classes at their root during software development. Rather than pursuing endless patching cycles, the agency contends that developers can prevent the weaknesses most likely to be exploited by threat actors by addressing fundamental causes in the development process.
Frontier artificial intelligence models like Claude are now being leveraged to identify and exploit vulnerabilities in programmable logic controllers (PLCs) that control critical infrastructure such as water utilities. This development demonstrates how advanced AI systems can be weaponized to streamline the reconnaissance and attack planning phases against operational technology environments.
Security researcher Chaotic Eclipse has released PrettyPrague, a proof-of-concept exploit for a previously unknown privilege escalation vulnerability in GenDigital Avast Antivirus. The zero-day flaw allows attackers to elevate privileges on affected systems running the antivirus software.
Unauthenticated attackers are actively exploiting a critical vulnerability in Langflow that enables remote arbitrary Python code execution. Tracked as CVE-2026-0768, the flaw represents a significant risk to deployments lacking proper access controls.