KidsNanny is a two-stage multimodal content moderation pipeline that achieves high accuracy and efficiency in detecting child safety threats, particularly excelling in text-embedded content.

This paper proposes a 3D CNN detector that leverages temporal artifacts to accurately identify high-quality deepfake videos, demonstrating robust detection even after social media re-encoding.

The paper introduces SEED, a large-scale benchmark dataset for tracing sequential deepfake facial edits, and proposes FAITH, a frequency-aware Transformer model that effectively detects and orders the…

BuddyBench introduces a novel, privacy-constrained multi-task benchmark that integrates longitudinal learning trajectories, standardized clinical assessments, and randomized trial data to advance pedi…

The paper proposes EMSFD, an evidence-based decision modeling approach that enhances synthetic face detection reliability and generalizability by explicitly modeling class evidence and incorporating u…

The paper analyzes current and proposed age verification methods and proposes an alternative using open standards and cryptography to achieve secure, privacy-preserving age checking.

The paper introduces ChildEval, a large-scale benchmark designed to systematically evaluate how well large language models can infer and follow complex, child-specific preferences during long-context…

The paper introduces PHTV-Scout, a novel framework that analyzes Douyin and Kwai data, revealing a high prevalence of potentially harmful teen videos, particularly CSE imagery, and demonstrating that…

The paper proposes a privacy-preserving system for crowd monitoring that counts individuals across different locations and time periods using face recognition without ever revealing personal identitie…

The paper audits six LLMs across four languages, finding that their gender stereotyping is significantly wider than human baselines and that cross-lingual translation fundamentally alters the nature o…

The paper introduces 'contrastive privacy,' a formal, model-agnostic, and quantitative method for evaluating the semantic success of AI-based sanitization across multiple media modalities.

The paper demonstrates that supervised fine-tuning significantly outperforms frontier zero-shot large language models for screen-conditioned action prediction on the PiSAR benchmark, highlighting the…

The paper argues that deepfake detection research is misaligned because it focuses on historical threats (public-figure face-swaps) while ignoring the dominant, emerging harms like NCII, voice-cloning…

The paper proposes a sequence-alignment framework using Soft Dynamic Time Warping to evaluate audio-driven talking-head generation, demonstrating that this approach provides more robust and fair compa…

The paper demonstrates that using synthetic hand images containing accessories, generated via inpainting, significantly improves the robustness of hand detectors for safety-critical applications by cl…

The paper proposes a novel, practical upper bound to estimate the worst-case performance of medical prediction models on the target population, even when the selection bias mechanism and target data a…

The paper introduces ImageProtector, a user-side method that embeds an imperceptible perturbation into images to prevent Multi-modal Large Language Models (MLLMs) from analyzing and extracting sensiti…

The paper introduces FraudBench, a multimodal benchmark designed to detect AI-generated fraudulent refund evidence, finding that current AI models struggle significantly with claim-conditioned fake-da…