Five US federal agencies have dropped a joint advisory that should stop every industrial control engineer in their tracks. The NSA, CISA, FBI, Department of Energy, and Environmental Protection Agency warned this week that threat actors are actively using artificial intelligence to write exploit scripts targeting internet-exposed Siemens S7 programmable logic controllers across water, energy, manufacturing, and other critical infrastructure sectors.
This is not a drill. The agencies were explicit: “this is not a theoretical risk, it is an active threat.”
What the attackers are doing
The advisory details a disturbing evolution in capability. Threat actors are combining open-source industrial automation libraries, specifically snap7.dll and python-snap7, with AI-assisted scripting to create custom tools that mimic legitimate operational technology monitoring solutions. These tools provide read and write access to Siemens S7 Series PLC memory, configuration data, and ladder logic programs via the S7comm protocol.
Here is the part that should concern you: AI is dramatically reducing both the technical expertise and time required to develop working industrial control system exploitation scripts. An attacker who previously needed deep knowledge of PLC architectures and industrial protocols can now point an AI model at a library and get a functioning exploit in minutes. The agencies note that AI also lets adversaries rapidly leverage additional attack vectors and adapt to defensive measures.
Who is in the crosshairs
The affected Siemens product lines are broad: S7-200, S7-300, S7-400, S7-1200, and S7-1500, including the F-series safety controllers. The sectors most targeted are critical manufacturing, energy, water and wastewater, chemical, food and agriculture, and commercial facilities.
The attack pattern follows a familiar reconnaissance-to-disruption playbook. Attackers scan the internet with tools like Censys and ZoomEye to find exposed or poorly segmented Siemens S7 devices. They then exploit default or weakly configured credentials to gain access. Once inside, they use read access to understand the target environment, preparing for future write operations that could cause physical disruption.
What organisations should do right now
The advisory’s recommended mitigations are straightforward but require immediate action:
- Inventory every Siemens S7 device on the network
- Apply all available security patches
- Keep PLCs off the internet entirely
- Strengthen access controls and remove default credentials
- Monitor for unauthorised S7comm activity
- Harden PLC services and protocols
- Hunt for signs of compromise
If your organisation relies on system integrators or third-party service providers, share this advisory with them immediately. An asset owner may not know their PLCs are reachable from the internet until it is too late.
Why this changes the calculus
For years, industrial control systems operated under a reasonable assumption: the barrier to entry for a sophisticated OT attack was high. You needed specialist knowledge, physical access, or both. AI collapses that barrier. The same technology that helps defenders detect threats is now enabling less-skilled attackers to launch credible operations against critical infrastructure.
This advisory follows a pattern of recent AI-assisted incidents. In April, CISA warned that Iranian-affiliated actors were exploiting internet-connected Rockwell Automation PLCs. A July update expanded that scope to Schneider Electric and Siemens devices. Days later, a coordinated cyberattack hit operational technology systems at more than 30 community water utilities across Minnesota. Security researchers link that intrusion to the Iran-linked group CyberAv3ngers.
The trajectory is clear. AI is not just augmenting defence; it is democratising offence. Every script kiddie with a cloud AI subscription and a targets list now has access to capabilities that once required nation-state resources.
The uncomfortable truth is that offensive AI tools are outpacing defensive capabilities. Organisations that relied on obscurity or specialist knowledge as a security layer need to rethink their entire architecture.
Philip Hall
What this means for your team
If you manage or oversee industrial systems, treat this advisory as a call to action, not background reading. The AI-generated exploit scripts circulating now may be crude, but they will improve rapidly. The window for getting ahead of this threat is narrow.
Start with the basics: network segmentation, credential hygiene, and an honest inventory of what is actually exposed to the internet. If you cannot answer the question “which of our PLCs are reachable from the public internet” within an hour, that is your starting point.
Related Reading
- Nine Out of Ten Companies Are Not Ready for AI-Driven Attacks
- Gold Eagle Lands: The US Just Built an AI-Cybersecurity War Room
- Frontier AI Agents Took Unauthorised Actions in UK Safety Tests
The views expressed on this site are my own and do not represent those of any current or former employer. Articles are based on publicly available information and are provided for general educational purposes.
