FeatureBleed: Inferring Private Enriched Attributes From Sparsity-Optimized AI Accelerators

• URL: http://arxiv.org/abs/2602.18304v1
• Date: Fri, 20 Feb 2026 16:01:16 GMT
• Title: FeatureBleed: Inferring Private Enriched Attributes From Sparsity-Optimized AI Accelerators
• Abstract summary: Backend enrichment is now widely deployed in sensitive domains such as product recommendation pipelines, healthcare, and finance.<n>This paper presents the first hardware-level backend retrieval data-stealing attack.<n>Our attack, FEATUREBLEED, exploits zero-skipping in AI accelerators to infer private backend-retrieved features solely through end-to-end timing.
• Score: 3.3465375629622494
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Backend enrichment is now widely deployed in sensitive domains such as product recommendation pipelines, healthcare, and finance, where models are trained on confidential data and retrieve private features whose values influence inference behavior while remaining hidden from the API caller. This paper presents the first hardware-level backend retrieval data-stealing attack, showing that accelerator optimizations designed for performance can directly undermine data confidentiality and bypass state-of-the-art privacy defenses. Our attack, FEATUREBLEED, exploits zero-skipping in AI accelerators to infer private backend-retrieved features solely through end-to-end timing, without relying on power analysis, DVFS manipulation, or shared-cache side channels. We evaluate FEATUREBLEED on three datasets spanning medical and non-medical domains: Texas-100X (clinical records), OrganAMNIST (medical imaging), and Census-19 (socioeconomic data). We further evaluate FEATUREBLEED across three hardware backends (Intel AVX, Intel AMX, and NVIDIA A100) and three model architectures (DNNs, CNNs, and hybrid CNN-MLP pipelines), demonstrating that the leakage generalizes across CPU and GPU accelerators, data modalities, and application domains, with an adversarial advantage of up to 98.87 percentage points. Finally, we identify the root cause of the leakage as sparsity-driven zero-skipping in modern hardware. We quantify the privacy-performance-power trade-off: disabling zero-skipping increases Intel AMX per-operation energy by up to 25 percent and incurs 100 percent performance overhead. We propose a padding-based defense that masks timing leakage by equalizing responses to the worst-case execution time, achieving protection with only 7.24 percent average performance overhead and no additional power cost.

Related papers

• Efficient and Privacy-Preserving Distribution Statistics Analytics on Mobile Spatial Data [76.32349627923843]
  Performing efficient distributed statistical analysis on spatial data is crucial for real-time mobile computing applications.<n>Existing privacy-preserving techniques often introduce excessive overhead or inaccuracies.<n>In this paper, we design, implement, and evaluate the first system that supports efficient and privacy-preserving distribution statistics analysis for mobile spatial data.
  arXiv  Detail & Related papers  (2026-05-25T12:37:27Z)
• Privacy on the Fly: A Predictive Adversarial Transformation Network for Mobile Sensor Data [32.36355095752335]
  We propose a real-time privacy-preserving framework that leverages historical signals to generate adversarial perturbations proactively.<n>Experiments on two datasets demonstrate that PATN substantially degrades the performance of privacy inference models.
  arXiv  Detail & Related papers  (2025-11-10T15:57:17Z)
• Differentially Private Synthetic Data Release for Topics API Outputs [63.79476766779742]
  We focus on one Privacy-Preserving Ads API: the Topics API, part of Google Chrome's Privacy Sandbox.<n>We generate a differentially-private dataset that closely matches the re-identification risk properties of the real Topics API data.<n>We hope this will enable external researchers to analyze the API in-depth and replicate prior and future work on a realistic large-scale dataset.
  arXiv  Detail & Related papers  (2025-06-30T13:46:57Z)
• FARFETCH'D: A Side-Channel Analysis Framework for Privacy Applications on Confidential Virtual Machines [26.74194304212783]
  Developers lack a systematic, efficient way to measure and compare leakage across real-world deployments.<n>We present FARFETCH'D, an open-source toolkit that offers side-channel tracing primitives on production AMD SEV-SNP hardware.<n>We show that FARFETCH'D pinpoints vulnerabilities and guides low-overhead mitigations based on oblivious memory and differential privacy.
  arXiv  Detail & Related papers  (2025-06-18T23:58:29Z)
• MDHP-Net: Detecting an Emerging Time-exciting Threat in IVN [42.74889568823579]
  We identify a new time-exciting threat model against in-vehicle network (IVN)<n>These attacks inject malicious messages that exhibit a time-exciting effect, gradually manipulating network traffic to disrupt vehicle operations and compromise safety-critical functions.<n>To detect time-exciting threat, we introduce MDHP-Net, leveraging Multi-Dimentional Hawkes Process (MDHP) and temporal and message-wise feature extracting structures.
  arXiv  Detail & Related papers  (2024-11-15T15:05:01Z)
• Optimizing Cross-Client Domain Coverage for Federated Instruction Tuning of Large Language Models [87.49293964617128]
  Federated domain-specific instruction tuning (FedDIT) for large language models (LLMs) aims to enhance performance in specialized domains using distributed private and limited data.<n>We empirically establish that cross-client domain coverage, rather than data heterogeneity, is the pivotal factor.<n>We introduce FedDCA, an algorithm that explicitly maximizes this coverage through diversity-oriented client center selection and retrieval-based augmentation.
  arXiv  Detail & Related papers  (2024-09-30T09:34:31Z)
• Data-Oblivious ML Accelerators using Hardware Security Extensions [9.716425897388875]
  Outsourced computation can put client data confidentiality at risk. We develop Dolma, which applies DIFT to the Gemmini matrix multiplication accelerator. We show that Dolma incurs low overheads for large configurations.
  arXiv  Detail & Related papers  (2024-01-29T21:34:29Z)
• Toward Practical Privacy-Preserving Convolutional Neural Networks Exploiting Fully Homomorphic Encryption [11.706881389387242]
  Homomorphic encryption (FHE) is a viable approach for achieving private inference (PI) FHE implementation of a CNN faces significant hurdles, primarily due to FHE's substantial computational and memory overhead. We propose a set of optimizations, which includes GPU/ASIC acceleration, an efficient activation function, and an optimized packing scheme.
  arXiv  Detail & Related papers  (2023-10-25T10:24:35Z)
• TeD-SPAD: Temporal Distinctiveness for Self-supervised Privacy-preservation for video Anomaly Detection [59.04634695294402]
  Video anomaly detection (VAD) without human monitoring is a complex computer vision task. Privacy leakage in VAD allows models to pick up and amplify unnecessary biases related to people's personal information. We propose TeD-SPAD, a privacy-aware video anomaly detection framework that destroys visual private information in a self-supervised manner.
  arXiv  Detail & Related papers  (2023-08-21T22:42:55Z)
• G$^2$uardFL: Safeguarding Federated Learning Against Backdoor Attacks through Attributed Client Graph Clustering [116.4277292854053]
  Federated Learning (FL) offers collaborative model training without data sharing. FL is vulnerable to backdoor attacks, where poisoned model weights lead to compromised system integrity. We present G$2$uardFL, a protective framework that reinterprets the identification of malicious clients as an attributed graph clustering problem.
  arXiv  Detail & Related papers  (2023-06-08T07:15:04Z)
• Avoid Adversarial Adaption in Federated Learning by Multi-Metric Investigations [55.2480439325792]
  Federated Learning (FL) facilitates decentralized machine learning model training, preserving data privacy, lowering communication costs, and boosting model performance through diversified data sources. FL faces vulnerabilities such as poisoning attacks, undermining model integrity with both untargeted performance degradation and targeted backdoor attacks. We define a new notion of strong adaptive adversaries, capable of adapting to multiple objectives simultaneously. MESAS is the first defense robust against strong adaptive adversaries, effective in real-world data scenarios, with an average overhead of just 24.37 seconds.
  arXiv  Detail & Related papers  (2023-06-06T11:44:42Z)
• Over-the-Air Federated Learning with Privacy Protection via Correlated Additive Perturbations [57.20885629270732]
  We consider privacy aspects of wireless federated learning with Over-the-Air (OtA) transmission of gradient updates from multiple users/agents to an edge server. Traditional perturbation-based methods provide privacy protection while sacrificing the training accuracy. In this work, we aim at minimizing privacy leakage to the adversary and the degradation of model accuracy at the edge server.
  arXiv  Detail & Related papers  (2022-10-05T13:13:35Z)
• A Privacy-Preserving-Oriented DNN Pruning and Mobile Acceleration Framework [56.57225686288006]
  Weight pruning of deep neural networks (DNNs) has been proposed to satisfy the limited storage and computing capability of mobile edge devices. Previous pruning methods mainly focus on reducing the model size and/or improving performance without considering the privacy of user data. We propose a privacy-preserving-oriented pruning and mobile acceleration framework that does not require the private training dataset.
  arXiv  Detail & Related papers  (2020-03-13T23:52:03Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.