Hackers Could Inflate EV Charging Bills by Manipulating User Data, New Study Warns
As electric vehicles (EVs) continue to gain mainstream adoption, the infrastructure supporting them—especially public charging networks—has become a critical component of modern transportation. Behind the scenes, sophisticated energy management systems (EMS) orchestrate when and how EVs charge, aiming to balance user convenience with cost efficiency for operators. However, a groundbreaking new study reveals a previously overlooked vulnerability: cyber attackers could exploit user data to artificially inflate charging costs, potentially causing significant financial harm to charging station operators.
The research, conducted by Ye Chao and Cao Ning from Chongqing Vocational College of Applied Technology, introduces a novel cyberattack model that targets the very foundation of EV charging optimization—the user data submitted by drivers. Unlike traditional attacks that aim to disrupt service or steal credentials, this method is subtler and more insidious: it manipulates the input parameters that EMS uses to calculate optimal charging schedules, leading to higher electricity bills without triggering immediate alarms.
Published in Computing Technology and Automation, the study demonstrates how a malicious actor with access to the communication channel between an EV and a charging station could alter key pieces of information such as arrival time, departure time, initial state of charge (SOC), and desired final SOC. These seemingly minor adjustments, when processed by the station’s optimization algorithm, can result in charging plans that consume more energy or shift charging to peak-rate periods—both of which increase the total cost of operation.
What makes this threat particularly alarming is its stealth. The attack does not require breaching the core control systems of the charging station. Instead, it exploits the trust-based nature of data exchange in current EV charging protocols. When a driver approaches a charging station, their vehicle typically transmits a set of preferences and constraints—how long they plan to stay, how much charge they need, and so on. The EMS then uses this data to schedule charging at the lowest possible cost, often leveraging time-of-use electricity pricing to avoid expensive peak hours.
But if an attacker intercepts and slightly modifies this data—say, increasing the desired final SOC from 80% to 95%, or delaying the departure time by an hour—the EMS will dutifully recalculate the optimal charging plan based on the corrupted inputs. Since the system assumes the data is legitimate, it proceeds without raising any red flags. The result? A charging session that costs significantly more than it should, with the financial loss absorbed by the operator.
The researchers modeled this attack as a two-stage optimization problem. In the upper layer, the attacker determines the optimal set of data perturbations—how much to tweak each parameter—to maximize the charging cost while minimizing the risk of detection. In the lower layer, the EMS runs its standard cost-minimization algorithm using the manipulated data. By applying Karush-Kuhn-Tucker (KKT) conditions, the team transformed this complex bi-level model into a single-level optimization problem, making it computationally tractable and demonstrating its feasibility in real-world scenarios.
To test their model, the researchers simulated a virtual charging station equipped with 40 EVs and six charging units across three power levels: 50kW, 100kW, and 200kW. They used a 24-hour time horizon with 15-minute intervals and incorporated real-world time-of-use electricity pricing. Under normal conditions, the total charging cost came to approximately 1,287.85 yuan. However, when the attack model was activated, the cost rose to 1,377.11 yuan—a nearly 7% increase, amounting to hundreds of dollars in additional expenses over time.
The simulation revealed two primary attack strategies. First, attackers could increase the desired state of charge, forcing vehicles to draw more energy than necessary. In one scenario, the total energy demand rose from 183.28 kWh to 206.70 kWh. Second, attackers could shift charging sessions into higher-priced time blocks. For example, several EVs that would have charged during off-peak hours were rescheduled to charge during the 4 PM to 10 PM window, when electricity rates were significantly higher. In some cases, the charging window was moved earlier by as much as 45 minutes, directly increasing the cost.
Perhaps most concerning is that these changes were designed to stay within plausible bounds. The attackers did not set unrealistic departure times or demand impossibly high charge levels—tactics that might alert operators or users. Instead, they made small, incremental adjustments that remained within the expected range of normal user behavior. This makes the attack extremely difficult to detect using conventional anomaly detection systems.
The implications of this research extend beyond financial loss. As EV charging networks become more integrated with smart grids and demand-response programs, the integrity of user data becomes a cornerstone of grid stability. If attackers can manipulate charging patterns at scale, they could potentially destabilize local power distribution networks, create artificial demand spikes, or even interfere with renewable energy integration efforts.
Moreover, the study highlights a fundamental tension in the design of modern energy systems: the trade-off between automation and security. While automated optimization brings undeniable efficiency gains, it also creates new attack surfaces. Systems that rely heavily on user input—especially when that input is not rigorously verified—are vulnerable to what the researchers call “data poisoning” attacks.
This is not the first time cybersecurity concerns have been raised about EV charging infrastructure. Previous studies have examined risks such as eavesdropping, man-in-the-middle attacks, and denial-of-service incidents. However, most of these focus on availability and confidentiality. The work by Ye and Cao shifts the focus to data integrity—a more subtle but equally dangerous threat.
One of the key contributions of the study is its emphasis on the economic motivation behind such attacks. Unlike hackers who seek to cause chaos or steal data, the attackers modeled here are financially driven. By increasing the operational costs of charging stations, they could potentially extort operators, gain competitive advantage, or even manipulate energy markets. In a world where charging networks are increasingly operated by private companies competing for market share, such tactics could have far-reaching consequences.
The researchers also explored potential detection and mitigation strategies. One approach involves implementing a confirmation step: after the EMS receives user data, it sends a summary back to the user’s vehicle for verification. Only after the user confirms the details would charging begin. This simple feedback loop could prevent many forms of data manipulation, as the user would notice discrepancies between their intended settings and the received plan.
Another detection method relies on post-charging analysis. By comparing the expected energy consumption—based on the original user input—with the actual energy delivered, operators could identify suspicious discrepancies. For instance, if a vehicle requested a 50 kWh charge but received 65 kWh without user authorization, that could trigger an investigation. However, the study notes that this method may not catch all attacks, especially if the manipulated data remains within reasonable bounds.
A more robust solution may lie in cryptographic authentication and data integrity checks. By digitally signing user data at the vehicle level and verifying it at the charging station, operators could ensure that the information has not been altered in transit. This would require upgrades to existing communication protocols such as ISO 15118, which currently supports secure data exchange but is not universally implemented.
The study also raises questions about liability and insurance in the EV ecosystem. If a charging station suffers financial losses due to a cyberattack, who is responsible? The vehicle manufacturer? The charging network operator? The software provider? As the lines between hardware, software, and service blur, traditional liability frameworks may no longer apply. The researchers suggest that cybersecurity insurance could play a role in transferring risk, similar to how cyber insurance is used in other critical infrastructure sectors.
From a policy perspective, the findings underscore the need for stronger regulatory oversight of EV charging systems. Current standards focus heavily on electrical safety and interoperability, but cybersecurity remains a secondary concern. As this research shows, even minor data manipulations can have significant economic impacts. Regulators may need to mandate data validation mechanisms, regular security audits, and incident reporting requirements for charging network operators.
For EV owners, the study serves as a reminder that their vehicles are not just modes of transportation but nodes in a vast digital network. Every time they plug in, they are exchanging data that could be intercepted or altered. While the risk to individual users may be low—since the financial impact is borne by the operator—the broader implications for system reliability and trust are significant.
The research also opens new avenues for future work. For example, could similar attacks be launched against vehicle-to-grid (V2G) systems, where EVs feed energy back into the grid? In such scenarios, manipulated data could not only increase costs but also distort energy market signals. Could attackers coordinate multiple charging stations to create synchronized demand spikes? These are questions that demand further investigation.
Another area of interest is the potential for defensive optimization. Just as attackers use optimization models to maximize damage, operators could develop counter-strategies to minimize vulnerability. For instance, an EMS could be designed to detect subtle inconsistencies in user data—such as a sudden increase in desired SOC without a corresponding change in trip length—and flag them for review. Machine learning models trained on historical data could identify patterns indicative of manipulation.
The study’s methodology itself is noteworthy. By framing the attack as a mathematical optimization problem, the researchers provide a rigorous framework for understanding and quantifying the threat. This approach allows for precise measurement of attack effectiveness, detection difficulty, and economic impact—metrics that are essential for risk assessment and mitigation planning.
It is important to note that the attack described requires a certain level of sophistication. The attacker must not only gain access to the communication network but also understand the internal workings of the EMS optimization algorithm. They need to know the time-of-use pricing structure, the available charging infrastructure, and the constraints under which the system operates. This suggests that the most likely threat actors are not random hackers but rather well-resourced entities with specific economic motives.
Nevertheless, as EV adoption grows and charging networks expand, the attack surface will only increase. More charging stations, more connected vehicles, and more data exchanges mean more opportunities for exploitation. The industry must proactively address these risks before they are exploited at scale.
In conclusion, the research by Ye Chao and Cao Ning sheds light on a critical and underappreciated vulnerability in the EV charging ecosystem. By demonstrating how user data manipulation can lead to significant financial losses, the study calls for a reevaluation of current security practices. As the world transitions to electric mobility, ensuring the integrity of charging data must become a top priority. Without it, the promise of efficient, affordable, and sustainable transportation could be undermined by those who seek to profit from its weaknesses.
Ye Chao, Cao Ning, Chongqing Vocational College of Applied Technology, Computing Technology and Automation, DOI:10.16339/j.cnki.jsjsyzdh.202401028