As location data have been increasingly adopted in location-based advertising (LBA), revealing locations to untrusted service providers has raised severe privacy concerns. Recent studies propose obfuscation mechanisms built upon geo-indistinguishability (geo-IND) to provide formal privacy guarantee. Unfortunately, due to the high degree of spatiotemporal regularity in human mobility pattern, the privacy cost will be unacceptably high in this situation, leading to accurate inference of user real locations. In this study, we identify this privacy risk in LBA scenarios under long-term and multi-platform assumption. We demonstrate an attacker can infer 75%∼90% of top-1 locations within a range of only 200 meters. To address it, we propose PrivLocAd , a novel system which can provide longitudinal privacy guarantee. The novelty of PrivLocAd stems from a novel surrogate-based obfuscation, which generates multiple surrogate locations to improve the privacy-utility trade-off. In addition, two novel obfuscation mechanisms, the two-stage Gaussian and multi-level surrogate generation mechanism in charge of surrogate generation can achieve the longitudinal privacy guarantee in intra- and inter-platform condition respectively. Our experimental results demonstrate PrivLocAd is able to defend against the attack, which reduces the inference rate to less than 1% of user top-1 locations in the 200 meter range.

Digital ad fraud continuously evolves through device spoofing, behavior simulation, and coordinated traffic attacks, making the detection of sophisticated fraudulent activities a persistent challenge. In real-world settings, only a small fraction of fraudulent traffic receives reliable labels, while vast amounts of unlabeled data may still contain latent fraud. This severely limits the generalization ability of conventional supervised learning models. To address this issue, we propose DiffVC—a novel contrastive learning framework tailored for ad fraud detection under the Positive-Unlabeled (PU) learning paradigm. DiffVC employs a multi-granularity diffusion augmentation strategy that builds upon a denoising diffusion probabilistic model to generate semantically augmented samples at three granularities: weak, medium, and strong. This strategy expands the latent fraud feature space while preserving diverse semantic information. We further introduce a diffusion-distance-calibrated similarity that dynamically adjusts constraints between augmented samples based on their diffusion distance, thereby improving the discrimination of unlabeled fraud. In addition, we design a dual-gated residual Transformer encoder that adaptively captures high-order feature interactions via gated residual connections and a channel recalibration mechanism. Experimental results on multiple real-world ad fraud detection datasets demonstrate that DiffVC consistently outperforms state-of-the-art methods and achieves stronger generalization. Our results confirm the effectiveness and practical applicability of DiffVC for ad fraud detection under label-scarce conditions.