Recently, autonomous agents built on large language models (LLMs) have experienced significant development and are being deployed in real-world applications.Through the usage of tools, these systems can perform actions in the real world.Given the agents' practical applications and ability to execute consequential actions, such autonomous systems can cause more severe damage than a standalone LLM if compromised.While some existing research has explored harmful actions by LLM agents, our study approaches the vulnerability from a different perspective.We introduce a new type of attack that causes malfunctions by misleading the agent into executing repetitive or irrelevant actions.Our experiments reveal that these attacks can induce failure rates exceeding 80% in multiple scenarios.Through attacks on implemented and deployable agents in multi-agent scenarios, we accentuate the realistic risks associated with these vulnerabilities.To mitigate such attacks, we propose self-examination defense methods.Our findings indicate these attacks are more difficult to detect compared to previous overtly harmful attacks, highlighting the substantial risks associated with this vulnerability.

Generative models have garnered significant interest in the realm of machine learning but are costly to produce and face growing regulatory constraints, requiring resource-heavy training and collaboration with various stakeholders, especially data providers. Such collaborative environments have given rise to a new threat known as model hijacking attacks. Adversaries can tamper with the training process to embed a hidden task, so that train/hijack high-end models at minimal costs or even sidestep regulations. In this paper, we extend the scope of model hijacking from classifiers to generative models by introducing the first model hijacking attack tailored for Generative Adversarial Networks (GANs), namely Neeko. Neeko is based on a novel U-Net-based Disguiser and allows a compromised GAN to generate authentic-looking images from its original distribution, but when downscaled, these images are visually changed to be from the hijacking dataset distribution. Through experiments on different image benchmark datasets, we demonstrate the efficacy and stealthiness of Neeko. Neeko poses security and accountability risks associated with training public GANs on potentially malicious or illegal datasets and raises concerns about evading those regulations addressing deepfakes and synthetic images.

Reasoning-focused LLMs sometimes alter their behavior when they detect that they are being evaluated—which can lead them to optimize for test-passing performance or to comply more readily with harmful prompts if real-world consequences appear absent. We present the first quantitative study of how such “test awareness” impacts model behavior, particularly its performance on safety-related tasks. We introduce a white-box probing framework that (i) linearly identifies awareness-related activations and (ii) steers models toward or away from test awareness while monitoring downstream performance. We apply our method to different state-of-the-art open-weight reasoning LLMs across both realistic and hypothetical tasks (denoting tests or simulations). Our results demonstrate that test awareness significantly impacts safety alignment (such as compliance with harmful requests and conforming to stereotypes) with effects varying in both magnitude and direction across models. By providing control over this latent effect, our work aims to provide a stress-test mechanism and increase trust in how we perform safety evaluations.

Generative Adversarial Networks (GANs) have become widely used in model training, as they can improve performance and/or protect sensitive information by generating data. However, this also raises potential risks, as malicious GANs may compromise or sabotage models by poisoning their training data. Therefore, it is important to verify the origin of a model’s training data for accountability purposes. In this work, we take the first step in the forensic analysis of models trained on GAN-generated data. Specifically, we first detect whether a model is trained on GAN-generated or real data. We then attribute these models, trained on GAN-generated data, to their respective source GANs. We conduct extensive experiments on three datasets, using four popular GAN architectures and four common model architectures. Empirical results show the remarkable performance of our detection and attribution methods. Furthermore, we conduct a more in-depth study and reveal that models trained on various data sources exhibit different decision boundaries and behaviours.

Large language model systems face significant security risks from maliciously crafted messages that aim to overwrite the system's original instructions or leak private data. To study this problem, we organized a capture-the-flag competition at IEEE SaTML 2024, where the flag is a secret string in the LLM system prompt. The competition was organized in two phases. In the first phase, teams developed defenses to prevent the model from leaking the secret. During the second phase, teams were challenged to extract the secrets hidden for defenses proposed by the other teams. This report summarizes the main insights from the competition. Notably, we found that all defenses were bypassed at least once, highlighting the difficulty of designing a successful defense and the necessity for additional research to protect LLM systems. To foster future research in this direction, we compiled a dataset with over 137k multi-turn attack chats and open-sourced the platform.

Algorithms such as Differentially Private SGD enable training machine learning models with formal privacy guarantees. However, because these guarantees hold with respect to unrealistic adversaries, the protection afforded against practical attacks is typically much better. An emerging strand of work empirically estimates the protection afforded by differentially private training as a confidence interval for the privacy budget $\hat{\varepsilon}$ spent with respect to specific threat models. Existing approaches derive confidence intervals for $\hat{\varepsilon}$ from confidence intervals for false positive and false negative rates of membership inference attacks, which requires training an impractically large number of models to get intervals that can be acted upon. We propose a novel, more efficient Bayesian approach that brings privacy estimates within the reach of practitioners. Our approach reduces sample size by computing a posterior for $\hat{\varepsilon}$ (not just a confidence interval) from the joint posterior of the false positive and false negative rates of membership inference attacks. We implement an end-to-end system for privacy estimation that integrates our approach and state-of-the-art membership inference attacks, and evaluate it on text and vision classification tasks. For the same number of samples, we see a reduction in interval width of up to 40% compared to prior work.

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 received less attention, which can be attributed to the false assumption that dataset curation techniques such as scrubbing are sufficient to prevent PII leakage. Scrubbing techniques reduce but do not prevent the risk of PII leakage: in practice scrubbing is imperfect and must balance the trade-off between minimizing disclosure and preserving the utility of the dataset. On the other hand, it is unclear to which extent algorithmic defenses such as differential privacy, designed to guarantee sentence-or user-level privacy, prevent PII disclosure. In this work, we introduce rigorous game-based definitions for three types of PII leakage via black-box extraction, inference, and reconstruction attacks with only API access to an LM. We empirically evaluate the attacks against GPT-2 models fine-tuned with and without defenses in three domains: case law, health care, and e-mails. Our main contributions are (i) novel attacks that can extract up to 10× more PII sequences than existing attacks, (ii) showing that sentence-level differential privacy reduces the risk of PII disclosure but still leaks about 3% of PII sequences, and (iii) a subtle connection between record-level membership inference and PII reconstruction. Code to reproduce all experiments in the paper is available at https://github.com/microsoft/analysing_pii_leakage.

Deploying machine learning models in production may allow adversaries to infer sensitive information about training data. There is a vast literature analyzing different types of inference risks, ranging from membership inference to reconstruction attacks. Inspired by the success of games (i.e. probabilistic experiments) to study security properties in cryptography, some authors describe privacy inference risks in machine learning using a similar game-based style. However, adversary capabilities and goals are often stated in subtly different ways from one presentation to the other, which makes it hard to relate and compose results. In this paper, we present a game-based framework to systematize the body of knowledge on privacy inference risks in machine learning. We use this framework to (1) provide a unifying structure for definitions of inference risks, (2) formally establish known relations among definitions, and (3) to uncover hitherto unknown relations that would have been difficult to spot otherwise.

Machine learning (ML) has made tremendous progress during the past decade and is being adopted in various critical real-world applications. However, recent research has shown that ML models are vulnerable to multiple security and privacy attacks. In particular, backdoor attacks against ML models have recently raised a lot of awareness. A successful backdoor attack can cause severe consequences, such as allowing an adversary to bypass critical authentication systems. Current backdooring techniques rely on adding static triggers (with fixed patterns and locations) on ML model inputs which are prone to detection by the current backdoor detection mechanisms. In this paper, we propose the first class of dynamic backdooring techniques against deep neural networks (DNN), namely Random Backdoor, Backdoor Generating Network (BaN), and conditional Backdoor Generating Network (c-BaN). Triggers generated by our techniques can have random patterns and locations, which reduce the efficacy of the current backdoor detection mechanisms. In particular, BaN and c-BaN based on a novel generative network are the first two schemes that algorithmically generate triggers. Moreover, c-BaN is the first conditional backdooring technique that given a target label, it can generate a target-specific trigger. Both BaN and c-BaN are essentially a general framework which renders the adversary the flexibility for further customizing backdoor attacks. We extensively evaluate our techniques on three benchmark datasets: MNIST, CelebA, and CIFAR-10. Our techniques achieve almost perfect attack performance on back-doored data with a negligible utility loss. We further show that our techniques can bypass current state-of-the-art defense mechanisms against backdoor attacks, including ABS, Februus, MNTD, Neural Cleanse, and STRIP.

Deep neural networks (DNNs) have progressed rapidly during the past decade and have been deployed in various real-world applications. Meanwhile, DNN models have been shown to be vulnerable to security and privacy attacks. One such attack that has attracted a great deal of attention recently is the backdoor attack. Specifically, the adversary poisons the target model’s training set to mislead any input with an added secret trigger to a target class.

In a membership inference attack, an attacker aims to infer whether a data sample is in a target classifier's training dataset or not. Specifically, given a black-box access to the target classifier, the attacker trains a binary classifier, which takes a data sample's confidence score vector predicted by the target classifier as an input and predicts the data sample to be a member or non-member of the target classifier's training dataset. Membership inference attacks pose severe privacy and security threats to the training dataset. Most existing defenses leverage differential privacy when training the target classifier or regularize the training process of the target classifier. These defenses suffer from two key limitations: 1) they do not have formal utility-loss guarantees of the confidence score vectors, and 2) they achieve suboptimal privacy-utility tradeoffs. In this work, we propose MemGuard,the first defense with formal utility-loss guarantees against black-box membership inference attacks. Instead of tampering the training process of the target classifier, MemGuard adds noise to each confidence score vector predicted by the target classifier. Our key observation is that attacker uses a classifier to predict member or non-member and classifier is vulnerable to adversarial examples.Based on the observation, we propose to add a carefully crafted noise vector to a confidence score vector to turn it into an adversarial example that misleads the attacker's classifier. Specifically, MemGuard works in two phases. In Phase I, MemGuard finds a carefully crafted noise vector that can turn a confidence score vector into an adversarial example, which is likely to mislead the attacker's classifier to make a random guessing at member or non-member. We find such carefully crafted noise vector via a new method that we design to incorporate the unique utility-loss constraints on the noise vector. In Phase II, MemGuard adds the noise vector to the confidence score vector with a certain probability, which is selected to satisfy a given utility-loss budget on the confidence score vector. Our experimental results on three datasets show that MemGuard can effectively defend against membership inference attacks and achieve better privacy-utility tradeoffs than existing defenses. Our work is the first one to show that adversarial examples can be used as defensive mechanisms to defend against membership inference attacks.

Abstract The decreasing costs of molecular profiling have fueled the biomedical research community with a plethora of new types of biomedical data, enabling a breakthrough towards more precise and personalized medicine. Naturally, the increasing availability of data also enables physicians to compare patients’ data and treatments easily and to find similar patients in order to propose the optimal therapy. Such similar patient queries (SPQs) are of utmost importance to medical practice and will be relied upon in future health information exchange systems. While privacy-preserving solutions have been previously studied, those are limited to genomic data, ignoring the different newly available types of biomedical data. In this paper, we propose new cryptographic techniques for finding similar patients in a privacy-preserving manner with various types of biomedical data, including genomic, epigenomic and transcriptomic data as well as their combination. We design protocols for two of the most common similarity metrics in biomedicine: the Euclidean distance and Pearson correlation coefficient. Moreover, unlike previous approaches, we account for the fact that certain locations contribute differently to a given disease or phenotype by allowing to limit the query to the relevant locations and to assign them different weights. Our protocols are specifically designed to be highly efficient in terms of communication and bandwidth, requiring only one or two rounds of communication and thus enabling scalable parallel queries. We rigorously prove our protocols to be secure based on cryptographic games and instantiate our technique with three of the most important types of biomedical data – namely DNA, microRNA expression, and DNA methylation. Our experimental results show that our protocols can compute a similarity query over a typical number of positions against a database of 1,000 patients in a few seconds. Finally, we propose and formalize strategies to mitigate the threat of malicious users or hospitals.