U.S. Agencies Warn Iran-Linked Hackers Are Using AI-Generated Exploits to Target Water Systems Across 12 States

In late July, operators at a municipal water facility in Clayton County, Georgia, watched their automated monitoring systems unexpectedly go offline. The disruption forced a sudden shift to manual controls and triggered a precautionary boil-water advisory for 300,000 residents as pressure levels fluctuated across the network. The incident in Georgia was not an isolated mechanical failure. By mid-August, federal authorities confirmed that water and wastewater utilities across at least 12 states—including Minnesota, Michigan, South Dakota, and New Jersey—had experienced similar coordinated disruptions to their operational technology.[1][2]

The underlying mechanism driving these incidents was detailed on August 19, when a coalition of five U.S. agencies, including the Cybersecurity and Infrastructure Security Agency (CISA) and the National Security Agency, issued a joint advisory. The campaign specifically targets internet-exposed Siemens S7 Series programmable logic controllers (PLCs). These industrial computers serve as the critical nodes responsible for regulating physical processes across municipal infrastructure, translating digital commands into physical actions such as valve operation, water pressure regulation, and chemical treatment.[3][4][5]

What distinguishes this campaign from previous infrastructure probes is the integration of artificial intelligence into the attack chain. The federal advisory notes that threat actors are deploying AI-generated exploitation scripts that are carefully disguised as legitimate operational technology monitoring software. Rather than manually crafting custom malware for each target, the attackers are utilizing AI to rapidly generate code capable of gaining initial access, pilfering credentials, and executing denial-of-service commands against the controllers. This approach allows the threat actors to scale their operations across multiple facilities simultaneously.[3][5][7]

Historically, compromising a Siemens S7 controller required deep, specialized engineering knowledge of proprietary industrial protocols. By pairing AI-assisted scripting with open-source automation libraries, the attackers have fundamentally altered the required capability threshold. The AI tools automate the generation of code that can read and rewrite PLC memory and ladder logic, effectively bypassing the need for decades of specialized operational technology experience. This industrialization of exploit development means that less-skilled adversaries can now execute complex attacks that were previously the exclusive domain of elite state-sponsored hacking teams.[3][4][5]

While the federal advisory did not formally attribute the activity to a specific nation-state, multiple intelligence sources and cybersecurity experts have linked the campaign to Iran-nexus threat actors. This attribution aligns with a documented, multi-year pattern of Iranian operations targeting the U.S. water sector. In late 2023, the Islamic Revolutionary Guard Corps-linked group known as CyberAv3ngers executed a similar campaign against water facilities in Pennsylvania, compromising Israeli-made programmable logic controllers to force operators into manual control modes.[1][2][6]

The success of the current campaign relies less on sophisticated zero-day exploits than on systemic architectural weaknesses within municipal infrastructure. Attackers are utilizing automated internet scanning services, such as Censys and ZoomEye, to locate programmable logic controllers that are directly connected to the public internet. These exposed devices frequently operate with outdated software, lack basic firewall protections, or rely on default passwords, providing the AI-generated scripts with an unobstructed pathway into the facility's operational technology network.[3][4][5][6][7]

Despite the geographic breadth of the targeting, the physical consequences of the cyberattacks have remained strictly contained. No drinking water contamination has been reported in any of the affected states. Facility operators have successfully maintained system safety by severing network connections and reverting to manual overrides when digital interfaces are compromised. This resilience highlights a critical defense mechanism within the water sector: the ability of human operators to physically decouple mechanical processes from compromised digital networks before safety parameters are breached.[1][2][6]

The deployment of AI-generated scripts against industrial control systems represents a structural shift in operational technology risk that extends far beyond the water sector. Federal agencies are now directing critical infrastructure operators across the manufacturing, energy, and chemical sectors to immediately isolate all programmable logic controllers from the internet. The convergence of known vulnerabilities, accessible open-source exploitation libraries, and AI-assisted development creates a highly probable attack vector for any inadequately protected facility, necessitating a fundamental reevaluation of how municipal infrastructure is secured.[3][4][5][7]