Connected autonomous vehicles must accurately detect, and adhere, to traffic light signals to ensure safe and efficient traffic flow. Misinterpretation of traffic lights can result in potential safety issues for drivers and pedestrians. Recent work demonstrated attacks that projected structured light patterns onto vehicle cameras, causing traffic signs and traffic light color misinterpretation. In this work, we characterize a novel vulnerability of traffic light physical structures that can be exploited by attackers to deceive recognition systems. When visible and invisible laser light is projected onto traffic lights, it is scattered by its internal reflectors. To a vehicle’s camera, the reflected light appears the same as a genuine light source, resulting in dangerous red and green traffic light status misclassifications. We evaluate our attack against three state-of-the-art traffic light recognition models and show successful misclassification up to 25 meters from the target traffic light. Furthermore, the attack succeeds both in daytime and nighttime conditions both in static and moving vehicle scenarios up to 10 Km/h speed. To mitigate this threat, we propose a detection system based on light texture patterns that achieve 100% TPR and 1.8% FPR in our real-world scenarios

Underwater data infrastructures offer natural cooling and enhanced physical security compared to terrestrial facilities, but are susceptible to acoustic injection attacks that can disrupt data integrity and availability. This work presents a comprehensive surveillance framework for localizing and tracking close-range adversarial acoustic sources targeting offshore infrastructures, particularly underwater data centers (UDCs). We propose a heterogeneous receiver configuration comprising a fixed hydrophone mounted on the facility and a mobile hydrophone deployed on a dedicated surveillance robot. While using enough arrays of static hydrophones covering large infrastructures is not feasible in practice, off-the-shelf approaches based on time difference of arrival (TDOA) and frequency difference of arrival (FDOA) filtering fail to generalize for this dynamic configuration. To address this, we formulate a Locus-Conditioned Maximum A-Posteriori (LC-MAP) scheme to generate acoustically informed and geometrically consistent priors, ensuring a physically plausible initial state for a joint TDOA-FDOA filtering. We integrate this into an unscented Kalman filtering (UKF) pipeline, which provides reliable convergence under nonlinearity and measurement noise. Extensive Monte Carlo analyses, Gazebo-based physics simulations, and field trials demonstrate that the proposed framework can reliably estimate the 3D position and velocity of an adversarial acoustic attack source in real time. It achieves sub-meter localization accuracy and over 90

Portable genome sequencers are revolutionizing genomic research. However, their reliance on external systems introduces new vulnerabilities that threaten the security of these sequencers. By employing zero-trust principles throughout the sequencing workflow, we can enhance the security of portable sequencing technology.

Ransomware attacks have become increasingly frequent and high-profile, resulting in billions of dollars in data and operational losses annually. Current mechanisms typically deploy defenses in vulnerable operating systems, making them susceptible to advanced adversaries capable of compromising the OS. While implementing defense mechanisms within storage devices can address this vulnerability, they lack detection accuracy due to their inability to access data semantics, such as file system metadata. Moreover, these methods only expose block-level interfaces without file-level information, limiting the usability and practicality of data recovery management. Therefore, we develop SrFTL, a novel ransomware defense framework that allows leveraging data semantics for accurate ransomware detection and effective file-level data recovery against data compromise. Specifically, SrFTL employs defense enforcement within the flash translation layer (FTL) of SSDs. Then, SrFTL combines the secure enclave with the modified FTL through a secure channel to enable flexible ransomware defenses within the enclave. Finally, SrFTL deploys ransomware classification and data recovery defenses in the enclave, providing high detection accuracy and low-cost data recovery. Our evaluation demonstrates that SrFTL achieves zero false positives and negatives when detecting our collected real-world ransomware samples and benign applications, outperforming current FTL-level solutions (e.g., MimosaFTL). Moreover, SrFTL introduces on average a trivial performance overhead of 1.5% compared with a regular SSD. Finally, evaluating against multiple real-world ransomware samples, SrFTL enables fast data recovery with an average time of 9.3 seconds. SrFTL thus bridges the semantic gap between the FTL and OS-level file information to stop ransomware while maintaining the integrity and authenticity of employed defenses.

Cyberattacks resulting in data loss remain a critical concern in modern data protection. To mitigate such threats, data versioning has been introduced to recover compromised data by reverting the storage to a prior uncompromised state. However, most current versioning solutions are implemented at the host level (e.g., within the operating system), making them vulnerable to adversaries with escalated privileges who can compromise OS-level protections. Thus, device-level methods have been proposed to shift the versioning logic to hardware-isolated storage devices outside the untrusted OS. Unfortunately, these solutions suffer from limited retention times for historical data, narrowing the protection window and leaving systems exposed to persistent attacks. In this paper, we propose LAST, an invaLidation-Aware VerSioning sysTem for flash-based SSDs, that enables data versioning with enhanced awareness of data retention time, ensuring long-term availability of historical data with small performance impact. LAST modifies the SSD's flash translation layer (FTL) to retain the data invalidation order for tracking data retention time. Then, it leverages an ordered garbage collection (GC) that always reclaims versioned data with the longest retention time, as determined by the invalidation sequence. Therefore, this approach prevents the premature deletion of data with shorter retention, significantly extending the protection window and reducing the risk of data loss. Evaluated under various real-world workloads, LAST achieves a small latency overhead of 1.5% over a regular SSD while maintaining data history for up to 126.4 days with an average of 52.6 days. This significantly outperforms the average retention of current versioning methods by 61.4% at least and 165.9% at most, enhancing the protection window against data loss from cyberattacks.

Plausible deniability (PD) allows at-risk users to deny the existence of their sensitive data stored on storage devices. This is critical to protect the privacy and the personal safety of users, as adversaries might force users to decrypt their devices, risking the disclosure of sensitive data that could endanger their lives and liberty. In this work, we show how current PD systems built on flash memory fail to obscure distinguishable data layouts created when hidden data is written. This deficiency makes them vulnerable to coercive adversaries who can capture single or multiple data snapshots of storage devices for scrutiny. To defend against this threat, we propose MUTE, a perMUTation-based PD systEm designed for flash memory. Building upon widely-adopted full disk encryption (FDE) mechanisms that provide device-level data encryption, MUTE modifies the distribution of initialization vectors (IVs) for encryption blocks within FDE, translating the hidden data into a permutation derived from the IV. Unlike other PD solutions, MUTE allows for storing hidden data without requiring the reduction of storage capacity. Moreover, it preserves the plausible deniability of the hidden data in a provably secure manner by maintaining the original logic of data operations on the flash memory without changing the data layout. We implement MUTE in the flash translation layer (FTL) of flash-based SSDs using FEMU, a widely-used emulator supporting flash memory research. Our evaluation with various micro-benchmarks and real-world workloads demonstrates that MUTE provides practical write and read throughputs of 23.4 MB/s and 15.7 MB/s and a capacity of 25.3 GB for hidden data in a 512 GB SSD, comparable with existing PD systems. MUTE achieves strong PD guarantees for flash-based devices against coercive adversaries, outperforming current PD systems.

We introduce a novel electromagnetic (EM) side-channel attack that allows for acoustic eavesdropping on electronic devices. This method specifically targets modern digital microelectromechanical systems (MEMS) microphones, which transmit captured audio via pulse-density modulation (PDM), that translate the analog sound signal into the density of output pulses in the digital domain. We discover that each harmonic of these digital pulses retains acoustic information, allowing the original audio to be retrieved through simple FM demodulation using standard radio receivers. An attacker can exploit this phenomenon to capture what the victim microphone hears remotely without installing malicious software or tampering with the device. We verify the vulnerability presence by conducting real-world evaluation on several PDM microphones and electronic devices, including laptops and smart speakers. For example, we demonstrate that the attack achieves up to 94.2% accuracy in recognizing spoken digits, up to 2 meters from a victim laptop located behind a 25 cm concrete wall. We also evaluate the attacker capability to eavesdrop on speech using popular speech-to-text APIs (e.g., OpenAI) not trained on EM traces, achieving a maximum of 14% transcription error rate in recovering the Harvard Sentences dataset. We further demonstrate that similar accuracy can be achieved with a cheap and stealthy antenna made out of copper tape. We finally discuss the limited effectiveness of current defenses such as resampling, and we propose a new hardware defense based on clock randomization.

Recapturing computer monitors with a camera is a common threat to cryptographic techniques designed to verify the origin of images and prevent AI-generated deepfakes. Although depth information can help distinguish a real-world scene from a flat computer monitor, incorporating additional depth sensors is often cost-prohibitive. To address this challenge, we explore the use of dual-pixel (DP) image sensors commonly found in still and smartphone cameras for fast autofocus, as a means to extract depth information for distinguishing real scenes from recaptured ones, without requiring additional hardware. Our signal processing pipeline is composed of (i) a stereo matching algorithm to obtain a disparity map using a pair of images generated from a DP image sensor and (ii) plane fitting to evaluate the flatness of the scene. Our proof-of-concept evaluation on a real-world DP image dataset demonstrates that the proposed method detects recaptured images at 100% accuracy. Similarly, it successfully distinguishes real-world scenes from recaptured deepfake images with >98% accuracy.

Reproducible research ensures a foundation for credible and progressive scientific advancement. Calls for reproducibility in the computer security and privacy community resulted in ACM WiSec and ACSAC in 2017 creating artifact evaluation committees (AECs) to foster more reproducible outcomes. Upon acceptance to ACM WiSec and ACSAC, authors may submit experimental artifacts to the committee to demonstrate that their artifacts effectively support their paper. While this may seem to address reproducibility, it is unknown whether AECs have actually improved reproducibility. Furthermore, AECs only assess artifacts submitted to the committee, and thus, may not holistically reflect the status of reproducibility (e.g., artifacts that did not go through the AEC). We conduct an indirect (i.e., availability) and a direct (i.e., running code) reproducibility studies to measure and evaluate the potential and realized reproducibility by classifying and running artifacts associated with individual papers. We collect over 2,000 papers and 550 artifacts to measure the state of reproducibility within four established applied security conferences between 2013 and 2023, two of which have AECs and two comparable conferences without AECs. We find that only 10% of research provides runnable artifacts, and only 3.9% of published papers produce a reproducible artifact. More importantly, our work shows that including an AEC alone does not necessarily result in more reproducible artifacts; specifically, ACM WiSec artifacts are more or less in line with the two control conferences, whereas ACSAC ultimately achieves 90% of artifacts participating in the AEC, with 40% of artifacts running. As such, our work shows that while the security community is developing mechanisms to foster reproducible research, there are significant improvements needed to progress reproducibility within computer security and privacy research.

Embedded devices play a major role in supporting critical infrastructure, but lack many of the security pro-tections of sophisticated systems. Determining whether these devices are compromised is, therefore, a challenge. In this work, we describe a novel control flow verification methodology via electromagnetic (EM) emanations. We design a framework that incorporates signal processing and training to detect subtle control flow deviations as small as three clock cycles, the minimum required to execute a return with malicious activity on modern embedded hardware. Our methodology leverages basic block detection, enabling the discovery of these subtle control flow deviations that escape conventional detection approaches. We evaluate our framework's ability to detect insertion and modification control flow attacks on six different case studies of real-world critical operations and two processors featuring different architectures. Our results show 96.6% detection accuracy across all tested programs and attacks. Finally, we show the transferability of our methodology to different instances of our evaluated processors, reaching up to 98.7% convergence with our baseline models while requiring a third of the collected EM samples compared to standard retraining. In doing so, we reveal the feasibility of fine-grained EM-based control flow monitoring for low-power microcontrollers.

Underwater Data Centers (UDC) offer natural cooling and built-in security for critical naval infrastructures. However, UDC storage devices are vulnerable to acoustic injection attacks as sound travels four times faster in water. In this work, we propose a subsea surveillance framework to identify and interpret acoustic attacks on UDCs in real-time. Our framework includes the system design of position error signal (PES)-based energy signature modeling, an integrated COMSOL simulation engine for comprehensive attack rehearsal, and a datadriven algorithm for spatial-temporal-frequency domain analyses to map such attacks. Specifically, we investigate various acoustic attack conditions on a real UDC testbed as well as in COMSOL simulation for PES-based energy signatures for comprehensive data-driven modeling. We demonstrate that classical approaches such as a K-means clustering model can then be trained to detect potential attacks in real-time from the energy signatures inside UDC pods. We are also in the process of deploying a full-size UDC pod to the ocean for more comprehensive field experimental validation. The subsea pod will ensure real operational scenarios and open up promising opportunities for future research on UDC surveillance.

Interpreting deep neural networks through concept-based explanations offers a bridge between low-level features and high-level human-understandable semantics. However, existing automatic concept discovery methods often fail to align these extracted concepts with the model’s true decision-making process, thereby compromising explanation faithfulness. In this work, we propose FACE (Faithful Automatic Concept Extraction), a novel framework that combines Non-negative Matrix Factorization (NMF) with a Kullback-Leibler (KL) divergence regularization term to ensure alignment between the model’s original and concept-based predictions. Unlike prior methods that operate solely on encoder activations, FACE incorporates classifier supervision during concept learning, enforcing predictive consistency and enabling faithful explanations. We provide theoretical guarantees showing that minimizing the KL divergence bounds the deviation in predictive distributions, thereby promoting faithful local linearity in the learned concept space. Systematic evaluations on ImageNet, COCO, and CelebA datasets demonstrate that FACE outperforms existing methods across faithfulness and sparsity metrics.

This paper studies the capability of light-based signal injection attacks to remotely inject fake inputs to modern optical sensor-based input devices. We demonstrate how an attacker can successfully inject malicious Character User Interface (CUI) commands on different laser-projection keyboards and generate fake mouse-cursor movements in bending mouses, using invisible to human eyes, infrared lasers. Our proof-of-concept evaluation shows how the attack achieves 100% success rate on injecting basic keyboard operations up to 7 meters away from the victim input device and through glass windows, without tampering with the victim computer system, or needing a network connection to pursue the attack. This vulnerability allows the attacker not only to type unauthorized commands but also to prevent the unlocked victim PC or workstation from automatically going to sleep mode, thereby extending the time window for locally committing malicious activities, i.e., lunchtime attacks. Within a wide viewing angle of 30°–45° from the victim input device location, the attacker can continuously inject false movements and press at least 24 consecutive keys without any failure up to 5 meters away. We also verify the attack feasibility in realistic office environments where the laser beam is partially occluded. Our analysis shows the potential security risks of invisible light injection attacks, including providing preventive defense measures to limit exposure of optical input interfaces to such a threat. This work aims to help manufacturers address the vulnerability and build reliable Human-Computer Interfaces.

Inertial sensors are widely used in navigation, motion tracking, and gesture recognition systems. However, these sensors are vulnerable to spoofing attacks, where an attacker injects a carefully designed acoustic signal to trick the sensor readings. Traditional approaches to detecting and mitigating attacks rely on module redundancy, i.e., adding multiple sensor modules to increase robustness. However, this approach is not always feasible due to the limited space and increased complexity of current printed circuit boards. This paper proposes a new method, ADC-Bank, to detect inertial sensor spoofing attacks via acoustic out-of-band signals. Unlike other multiple-sensor-based solutions, it is based on component redundancy within one sensor, using multiple analog-to-digital converters (ADCs) with different sampling rates to simultaneously sample the output of the sensors. The different sample rates result in different aliasing frequencies for out-of-band signals that can be used to detect attacks. The proposed method is evaluated on off-the-shelf inertial sensors with commercial ADCs, demonstrating its ability to detect the attacking signals with relatively low cost and computation overhead.

Underwater data centers (UDCs) hold promise as next-generation data storage due to their energy efficiency and environmental sustainability benefits. While the natural cooling properties of water save power, the isolated aquatic environment and long-range sound propagation characteristics in water create unique vulnerabilities which differ from those of on-land data centers. Our research discovers the unique vulnerabilities of fault-tolerant storage devices, resource allocation software, and distributed file systems to acoustic injection attacks in UDCs. With a realistic testbed approximating UDC server operations, we empirically characterize the capabilities of acoustic injection underwater and find that an attacker can reduce fault-tolerant RAID 5 storage system throughput by 17% up to 100%. Our closed-water analyses reveal that an attacker can (i) cause unresponsiveness and automatic node removal in a distributed filesystem with only 2.4 minutes of sustained acoustic injection, (ii) induce a distributed database's latency to increase by up to 92.7% to reduce system reliability, and (iii) induce load-balance managers to redirect up to 74% of resources to a target server to cause overload or force resource colocation. Furthermore, we perform open-water experiments in a lake and find that an attacker can cause controlled throughput degradation at the maximum allowable distance of 6.35 m using a commercial speaker. We also investigate and discuss the effectiveness of standard defenses against acoustic injection attacks. Finally, we formulate a novel machine learning-based detection system that reaches 0% False Positive Rate and 98.2% True Positive Rate trained on our dataset of profiled hard disk drives under 30-second FIO benchmark execution. With this work, we aim to help manufacturers proactively protect UDCs against acoustic injection attacks and ensure the security of subsea computing infrastructures.