Introduction
In the world of cyber-physical systems cybersecurity, 2025 was a landmark year—and one that many companies are likely eager to move past. Not only did attackers grow more targeted and precise in the scale and severity of their attacks against industrial and healthcare companies, but organizations also suffered historic losses as a result. And with the ongoing convergence of IT and CPS, CISOs have had to move quicker than ever to adapt and implement robust security programs that protect both.
In particular, attacks against CPS are on the rise. As many of those systems find their way online for the first time, the scale and frequency of attacks will show no signs of slowing down.
With all this in mind, here are five of our predictions for CPS security going into 2026.
Attacks are Coming to CPS
Threat actors are coming for cyber-physical systems.
We’re expecting an evolution in the way hackers target operational technology (OT), connected medical devices, and building management systems (BMS). Past research, for example, has demonstrated potential soft spots in devices living at certain levels of the Purdue Model. Vulnerabilities on engineering workstation software that largely operates on Windows machines were perceived as risky given that illicit access could enable an attacker to disrupt or manipulate data it was uploaded or downloaded from programmable logic controllers (PLCs) and other industrial control systems.
Now given the widespread connectivity on shop floors, hospitals, and smart buildings, attackers have many more options to target CPS. Expect threat actors—and researchers—to increase their attention on smart internet of things devices, smart building automation, even connected intercoms and video surveillance systems. Enhanced connectivity represents more potential entry points onto the network.
Threat actors—depending on their motivations—can exploit a new landscape of vulnerabilities for illicit access to these systems and impact in numerous ways from extortion-style ransomware attacks to disruptive attacks that impact environmental conditions inside of buildings. Many of these non-Windows-based smart systems are not monitored, nor do they support security systems such as endpoint detection and response (EDR), and present the type of black hole often coveted by a threat actor.
Attackers Living on the Edge
Speaking of EDR, the technology is ubiquitous on enterprise networks and is largely successful in detecting and blocking commodity attacks such as ransomware and exploits of known vulnerabilities. Edge devices, however, such as 4G routers that many factories, smart buildings, and hospitals use as internet gateways and must be exposed to the network, often do not support EDR.
In the past 12 months, attackers have developed headline-grabbing frameworks targeting edge devices, hypervisors, and other network infrastructure that does not generally support EDR. Attacks that are publicly known have involved exploits against not only zero-day vulnerabilities but also commodity flaws in order to backdoor these types of technology.
CPS, meanwhile, also figures to eventually become a target of these types of attacks given that many of the IoT sensors and BMS prevalent inside enterprises that make up the CPS ecosystem also don’t support EDR. Attackers can exploit access to CPS for lateral movement and threaten physical manufacturing processes in factories and patient care in hospitals. This appears likely to put pressure on organizations to rely on network and threat monitoring for CPS, as well as a host of compensating controls to make up for some of the patching difficulties present in both OT and healthcare environments. Controls such as network segmentation go a long way toward containing these types of attacks and building resilient systems.
CPS Maturity Means Remediation
Cyber-physical systems protection programs must mature as rapidly as the environments in which they operate. Connectivity and convergence have introduced new efficiencies and visibility into data never before seen. These two factors have also introduced new risks that must be remediated, and not just catalogued.
While we predict enhanced attention on the targeting of CPS from advanced attackers and lower-sophistication groups alike, it’s time to implement a fully baked protection program around CPS assets; an inventory and identification of risk will not be sufficient moving forward.
Enterprises must assess the risks posed by edge devices that don’t support endpoint detection and remediation, and remediate network-based assets such as routers, firewalls, IoT sensors, smart devices, and remote access systems whenever possible. Software patches and firmware updates for these edge assets must be prioritized when available. A strategy around the application of compensating controls, such as robust network segmentation, must be considered as a mitigation as well.
The AI-Driven Democratization of Hacktivism
In 2026, the barrier to entry for becoming a hacker attacking internet-exposed ICS is the lowest it has ever been. Generative AI tools are effectively arming low-sophistication actors and "script kiddies" with capabilities previously reserved for nation-states.
We predict a surge in hacktivist activity where attackers no longer require in-depth knowledge of industrial protocols; the AI provides it for them. This will lead to a high volume of disruptive attacks on critical infrastructure, forcing defenders to rely heavily on AI-driven detection to filter out the noise of these empowered, lower-tier adversaries.
The Pivot to ‘Continuous Vetting’ for Insiders
The "trusted insider" is becoming a critical volatility factor. Following a rise in incidents where initial access was sold or exploited by disgruntled employees, we predict that critical infrastructure organizations will be forced to overhaul their personnel security.
By 2026, standard pre-employment background checks will no longer suffice; we will see a regulatory and operational shift toward "continuous personnel vetting." Companies will increasingly treat human employees with the same zero trust scrutiny as devices, requiring ongoing behavioral validation to prevent the monetization of access credentials.
