Related papers: Understanding Secret Leakage Risks in Code LLMs: A…

Standard Byte-Pair Encoding (BPE) tokenization compresses text by pairing a learned token vocabulary with a detailed merge list. Recent work has shown that this merge list exposes a potential attack surface for extracting information about…

Large language models (LLMs) have advanced natural language processing (NLP) skills such as through next-token prediction and self-attention, but their ability to integrate broad context also makes them prone to incorporating irrelevant…

Tokenization is the first -- and often least scrutinized -- step of most NLP pipelines. Standard algorithms for learning tokenizers rely on frequency-based objectives, which favor languages dominant in the training data and consequently…

Concept Bottleneck Models (CBMs) aim to enhance interpretability by structuring predictions around human-understandable concepts. However, unintended information leakage, where predictive signals bypass the concept bottleneck, compromises…

This work quantifies the risk of training data leakage from LLMs (Large Language Models) using sequence-level probabilities. Computing extraction probabilities for individual sequences provides finer-grained information than has been…

Large Language Models (LLMs) have demonstrated promising capabilities for code generation. While existing benchmarks evaluate the correctness and efficiency of LLM-generated code, the potential linguistic bias - where code quality varies…

The growing use of large language models in sensitive domains has exposed a critical weakness: the inability to ensure that private information can be permanently forgotten. Yet these systems still lack reliable mechanisms to guarantee that…

Large language models, trained on massive corpora, are prone to verbatim memorization of training data, creating significant privacy and copyright risks. While previous works have proposed various definitions for memorization, many exhibit…

Background: Leaking sensitive information - such as API keys, tokens, and credentials - in source code remains a persistent security threat. Traditional regex and entropy-based tools often generate high false positives due to limited…

Machine learning poses severe privacy concerns as it has been shown that the learned models can reveal sensitive information about their training data. Many works have investigated the effect of widely adopted data augmentation and…

Deployed large language models (LLMs) often rely on speculative decoding, a technique that generates and verifies multiple candidate tokens in parallel, to improve throughput and latency. In this work, we reveal a new side-channel whereby…

Training data memorization in NLP can both be beneficial (e.g., closed-book QA) and undesirable (personal data extraction). In any case, successful model training requires a non-trivial amount of memorization to store word spellings,…

Large language models (LLMs) are increasingly prevalent in security research. Their unique characteristics, however, introduce challenges that undermine established paradigms of reproducibility, rigor, and evaluation. Prior work has…

Fine-tuning on open-source Large Language Models (LLMs) with proprietary data is now a standard practice for downstream developers to obtain task-specific LLMs. Surprisingly, we reveal a new and concerning risk along with the practice: the…

Code generation aims to synthesize code and fulfill functional requirements based on natural language (NL) specifications, which can greatly improve development efficiency. In the era of large language models (LLMs), large code models…

State-of-the-art language models are autoregressive and operate on subword units known as tokens. Specifically, one must encode the conditioning string into a list of tokens before passing to the language models for next-token prediction.…

Language model (LM) agents that act on users' behalf for personal tasks (e.g., replying emails) can boost productivity, but are also susceptible to unintended privacy leakage risks. We present the first study on people's capacity to oversee…

Data encryption limits the power and efficiency of queries. Direct processing of encrypted data should ideally be possible to avoid the need for data decryption, processing, and re-encryption. It is vital to keep the data searchable and…

Large language models (LLMs) have shown remarkable proficiency in generating text, benefiting from extensive training on vast textual corpora. However, LLMs may also acquire unwanted behaviors from the diverse and sensitive nature of their…

In this paper, we analyse the generalization ability of binary classifiers for the task of deepfake detection. We find that the stumbling block to their generalization is caused by the unexpected learned identity representation on images.…