相关论文: Subject-level Inference for Realistic Text Anonymi…

Membership Inference Attacks (MIA) aim to infer whether a target data record has been utilized for model training or not. Existing MIAs designed for large language models (LLMs) can be bifurcated into two types: reference-free and…

Language Models (LMs) have been shown to leak information about training data through sentence-level membership inference and reconstruction attacks. Understanding the risk of LMs leaking Personally Identifiable Information (PII) has…

Defining privacy and related notions such as Personal Identifiable Information (PII) is a central notion in computer science and other fields. The theoretical, technological, and application aspects of PII require a framework that provides…

This work investigates and evaluates multiple defense strategies against property inference attacks (PIAs), a privacy attack against machine learning models. Given a trained machine learning model, PIAs aim to extract statistical properties…

AI chatbots have quietly become the world's most popular therapists, coaches, and confidants. Users of cloud-based LLM services are increasingly shifting from simple queries like idea generation and poem writing, to deeply personal…

The widespread usage of large-scale multimodal models like CLIP has heightened concerns about the leakage of PII. Existing methods for identity inference in CLIP models require querying the model with full PII, including textual…

Computer use agents create new privacy risks: training data collected from real websites inevitably contains sensitive information, and cloud-hosted inference exposes user screenshots. Detecting personally identifiable information in web…

The rise of chronic diseases and pandemics like COVID-19 has emphasized the need for effective patient data processing while ensuring privacy through anonymization and de-identification of protected health information (PHI). Anonymized data…

Meeting summarization with large language models (LLMs) remains error-prone, often producing outputs with hallucinations, omissions, and irrelevancies. We present FRAME, a modular pipeline that reframes summarization as a semantic…

Supervised Finetuning (SFT) has become one of the primary methods for adapting a large language model (LLM) with extensive pre-trained knowledge to domain-specific, instruction-following tasks. SFT datasets, composed of instruction-response…

Safety classifiers are essential safeguards within generative AI systems, filtering harmful content or identifying at-risk users when interacting with large language models. Despite their necessity, these models are trained on sensitive…

Current LLM-based frameworks for text anonymization usually rely on remote API services from powerful LLMs, which creates an inherent privacy paradox: users must disclose the raw data to untrusted third parties for guaranteed privacy…

Clinical personality assessment screens response validity before interpreting substantive scales. LLM evaluation does not. We apply the validity scaling framework from the PAI and MMPI-3 to metacognitive probe data from 20 frontier models…

Subject-driven text-to-image diffusion models have achieved remarkable success in preserving single identities, yet their ability to compose multiple interacting subjects remains largely unexplored and highly challenging. Existing…

We propose SETI (Systematicity Evaluation of Textual Inference), a novel and comprehensive benchmark designed for evaluating pre-trained language models (PLMs) for their systematicity capabilities in the domain of textual inference.…

Responsible use of AI demands that we protect sensitive information without undermining the usefulness of data, an imperative that has become acute in the age of large language models. We address this challenge with an on-premise,…

When different groups' values differ, one approach to model alignment is to steer models at inference time towards each group's preferences. However, techniques like in-context learning only consider similarity when drawing few-shot…

Anonymizing text that contains sensitive information is crucial for a wide range of applications. Existing techniques face the emerging challenges of the re-identification ability of large language models (LLMs), which have shown advanced…

Accurate recognition of specific categories, such as persons' names, dates or other identifiers is critical in many Automatic Speech Recognition (ASR) applications. As these categories represent personal information, ethical use of this…

Spear phishing is a deceptive attack that uses social engineering to obtain confidential information through targeted victimization. It is distinguished by its use of social cues and personalized information to target specific victims.…