Deepfake (DF) detectors face significant challenges when deployed in real-world environments, particularly when encountering test samples deviated from training data through either postprocessing manipulations or distribution shifts. We demonstrate postprocessing techniques can completely obscure generation artifacts presented in DF samples, leading to performance degradation of DF detectors. To address these challenges, we propose Think Twice before Adaptation (T2A), a novel online test-time adaptation method that enhances the adaptability of detectors during inference without requiring access to source training data or labels. Our key idea is to enable the model to explore alternative options through an Uncertainty-aware Negative Learning objective rather than solely relying on its initial predictions as commonly seen in entropy minimization (EM)-based approaches. We also introduce an Uncertain Sample Prioritization strategy and Gradients Masking technique to improve the adaptation by focusing on important samples and model parameters. Our theoretical analysis demonstrates that the proposed negative learning objective exhibits complementary behavior to EM, facilitating better adaptation capability. Empirically, our method achieves state-of-the-art results compared to existing test-time adaptation (TTA) approaches and significantly enhances the resilience and generalization of DF detectors during inference.

Deepfake (DF) detectors face significant challenges when deployed in real-world environments, particularly when encountering test samples deviated from training data through either postprocessingmanipulations or distribution shifts. We demonstrate postprocessing techniques can completely obscure generation artifacts presented in DF samples, leading to performance degradation of DF detectors. To address these challenges, we propose Think Twice before Adaptation (T2A), a novel online test-time adaptation method that enhances the adaptability of detectors during inference without requiring access to source training data or labels. Our key idea is to enable the model to explore alternative options through an Uncertainty-aware Negative Learning objective rather than solely relying on its initial predictions as commonly seen in entropy minimization (EM)-based approaches. We also introduce an Uncertain Sample Prioritization strategy and Gradients Masking technique to improve the adaptation by focusing on important samples and model parameters. Our theoretical analysis demonstrates that the proposed negative learning objective exhibits complementary behavior to EM, facilitating better adaptation capability. Empirically, our method achieves state-of-the-art results compared to existing test-time adaptation (TTA) approaches and significantly enhances the resilience and generalization of DF detectors during inference.

The advancements in automatic speaker recognition have led to the exploration of voice data for verification systems. This raises concerns about the security of storing voice templates in plaintext. In this paper, we propose a novel cancellable biometrics that does not require users to manage random matrices or tokens. First, we pre-process the raw voice data and feed it into a deep feature extraction module to obtain embeddings. Next, we propose a hashing scheme, Ranking-of-Elements, which generates compact hashed codes by recording the number of elements whose values are lower than that of a random element. This approach captures more information from smaller-valued elements and prevents the adversary from guessing the ranking value through Attacks via Record Multiplicity. Lastly, we introduce a fuzzy matching method, to mitigate the variations in templates resulting from environmental noise. We evaluate the performance and security of our method on two datasets: TIMIT and VoxCeleb1.

The rapid proliferation of generative AI has led to led to unprecedented capabilities in synthesizing realistic deepfakes (DFs) across multiple modalities. This raises significant concerns regarding privacy, security, and copyright protection. Unlike passive detection approaches that operate after DFs have been created and distributed, proactive defense mechanisms aim at preventing the generation of malicious synthetic content at its source. This article provides a comprehensive survey of current proactive DF defense strategies, including Disruption and Watermarking. Disruption approaches protect individuals data by introducing imperceptible perturbations that prevent unauthorized exploitation by generative models, while watermarking approaches embed verifiable messages into data or models to enable content authentication and attribution. We also analyze proactive approaches across various evaluation metrics (imperceptibility, protectability/detectability, transferability, traceability, and robustness), and examine their effectiveness in real-world settings. Furthermore, we review the evolution of DF generation techniques, highlighting their rapid developments. Finally, we identify key challenges and promising future research directions to enhance proactive defense mechanisms.

Neural networks unintentionally memorize training data, creating privacy risks in federated learning (FL) systems, such as inference and reconstruction attacks on sensitive data. To mitigate these risks and to comply with privacy regulations, Federated Unlearning (FU) has been introduced to enable participants in FL systems to remove their data’s influence from the global model. However, current FU methods primarily act post-hoc, struggling to efficiently erase information deeply memorized by neural networks. We argue that effective unlearning necessitates a paradigm shift: designing FL systems inherently amenable to forgetting. To this end, we propose a learning-to-unlearn Transformation-guided Federated Unlearning (ToFU) framework that incorporates transformations during the learning process to reduce memorization of specific instances. Our theoretical analysis reveals how transformation composition provably bounds instance-specific information, directly simplifying subsequent unlearning. Crucially, ToFU can work as a plug-and-play framework that improves the performance of existing FU methods. Experiments on CIFAR-10, CIFAR-100, and the MUFAC benchmark show that ToFU outperforms existing FU baselines, enhances performance when integrated with current methods, and reduces unlearning time.