AI agents aim to shield EV chargers from attacks

Researchers at the University of Malaga say EV chargers are becoming easier targets because they combine many physical and digital parts. Their proposal uses multiple AI agents—built around the Open Charge Point Protocol—to spot anomalies early, including frau

On a typical day, an EV charger might look like a simple piece of equipment—plug in, charge, move on. But the system behind it is anything but simple. Electric-vehicle charging stations bundle physical hardware with digital communications. and that layered design creates security exposure that researchers say hasn’t been studied deeply enough.

Cristina Alcaraz. an infrastructure-security researcher at Spain’s University of Malaga. points to the same problem in plain terms: charging stations carry liability because their complex architecture both keeps them running effectively and opens the door to “far-reaching security vulnerabilities.” When attackers can reach the chargers. the impact isn’t limited to one operator’s devices. It threatens continued EV adoption and adds pressure to the stability of electrical grids in the countries where chargers are deployed.

To address that risk. researchers from the NICS lab at the University of Malaga have developed a proposal to deploy AI agents designed to prevent cyberattacks coming from different directions—starting with fraud and energy theft carried out through charging stations. and stretching to larger attacks that could damage critical-energy networks.

The plan is built to work with the Open Charge Point Protocol. or OCPP. which the researchers say is one of the most widely used standards for operating and managing electric-vehicle chargers. OCPP lets a network of charging stations communicate with a centralized system that can manage. monitor. and coordinate energy transactions for end users.

That central system handles key functions remotely: user authentication. management of the electrical load at each station. monitoring overall electricity consumption. and technical diagnostics. For operators, these capabilities are meant to support real-time control—so anomalous behavior can be spotted and addressed quickly.

But the researchers argue the monitoring mechanisms in use today don’t go far enough. Current approaches based on OCPP tend to focus mainly on network traffic or local events. That narrow view makes it harder to determine where an anomaly is originating across a broader region of infrastructure. which specific network components have been compromised. how extensive vulnerabilities may be. and how a potential attack could spread.

The researchers’ proposal leans into that gap with a system of multiple AI agents. In their design. each station—or relevant component within the charging network—incorporates AI agents that can analyze what’s happening around them. collect information. and collaborate with other agents to form a fuller picture of the infrastructure’s present state.

“Each agent assesses the status of chargers. communications. and connected devices to detect anomalies. operational failures. or potential security incidents. ” Alcaraz says. “These agents. which are connected to a central-monitoring system. compare the information obtained locally with that of nearby stations. providing a more complete. accurate. and contextualized collaborative view of the situation. ” Alcaraz adds. She is also the lead author of the report.

One feature the study presents as especially novel is how it coordinates these agents. The work. published in the International Journal of Critical Infrastructure Protection. describes a consensus mechanism grounded in a mathematical framework called opinion dynamics. The idea is to mirror how humans exchange information across social networks to reach agreement. In the researchers’ model. AI agents share observations and then gradually adjust their assessments. building a collective understanding of what’s happening across the system.

Taken together, the proposal is aimed at one practical outcome: early and reliable detection of anomalies and attacks on charging networks—detection that operators can act on before the problem spreads beyond a single site.

AI agents EV chargers cybersecurity OCPP Open Charge Point Protocol critical infrastructure protection University of Malaga NICS lab opinion dynamics