Large Language Models (LLMs) are increasingly utilized in autonomous decision-making, where they sample options from vast action spaces.However, the heuristics that guide this sampling process remain under-explored.We study this sampling behavior and show that this underlying heuristics resembles that of human decision-making: comprising a descriptive component (reflecting statistical norm) and a prescriptive component (implicit ideal encoded in the LLM) of a concept.We show that this deviation of a sample from the statistical norm towards a prescriptive component consistently appears in concepts across diverse real-world domains like public health, and economic trends.To further illustrate the theory, we demonstrate that concept prototypes in LLMs are affected by prescriptive norms, similar to the concept of normality in humans.Through case studies and comparison with human studies, we illustrate that in real-world applications, the shift of samples toward an ideal value in LLMs' outputs can result in significantly biased decisionmaking, raising ethical concerns.

Large Language Models (LLMs) show impressive results in numerous practical applications, but they lack essential safety features that are common in other areas of computer science, particularly an explicit separation of instructions and data. This makes them vulnerable to manipulations such as indirect prompt injections and generally unsuitable for safety-critical tasks. Surprisingly, there is currently no established definition or benchmark to quantify this phenomenon. In this work, we close this gap by introducing a formal measure for instruction-data separation for single-turn language models and an empirical variant that is calculable from a model’s outputs. We also present a new dataset, SEP, that allows estimating the measure for real-world models. Our results on various LLMs show that the problem of instruction-data separation is real: all models fail to achieve high separation, and canonical mitigation techniques, such as prompt engineering and fine-tuning, either fail to substantially improve separation or reduce model utility.

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.

As the era of autonomous agents making decisions on behalf of users unfolds, ensuring contextual integrity (CI) -- what is the appropriate information to share while carrying out a certain task -- becomes a central question to the field. We posit that CI demands a form of reasoning where the agent needs to reason about the context in which it is operating. To test this, we first prompt LLMs to reason explicitly about CI when deciding what information to disclose. We then extend this approach by developing a reinforcement learning (RL) framework that further instills in models the reasoning necessary to achieve CI. Using a synthetic, automatically created, dataset of only $\sim700$ examples but with diverse contexts and information disclosure norms, we show that our method substantially reduces inappropriate information disclosure while maintaining task performance across multiple model sizes and families. Importantly, improvements transfer from this synthetic dataset to established CI benchmarks such as PrivacyLens that has human annotations and evaluates privacy leakage of AI assistants in actions and tool calls.

There is a growing interest in using Large Language Models (LLMs) in multi-agent systems to tackle interactive real-world tasks that require effective collaboration and assessing complex situations. Yet, we have a limited understanding of LLMs' communication and decision-making abilities in multi-agent setups. The fundamental task of negotiation spans many key features of communication, such as cooperation, competition, and manipulation potentials. Thus, we propose using scorable negotiation to evaluate LLMs. We create a testbed of complex multi-agent, multi-issue, and semantically rich negotiation games. To reach an agreement, agents must have strong arithmetic, inference, exploration, and planning capabilities while integrating them in a dynamic and multi-turn setup. We propose metrics to rigorously quantify agents' performance and alignment with the assigned role. We provide procedures to create new games and increase games' difficulty to have an evolving benchmark. Importantly, we evaluate critical safety aspects such as the interaction dynamics between agents influenced by greedy and adversarial players. Our benchmark is highly challenging; GPT-3.5 and small models mostly fail, and GPT-4 and SoTA large models (e.g., Llama-3 70b) still underperform in reaching agreement in non-cooperative and more difficult games.

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.

Clickbait PDFs are PDF documents that do not embed malware but trick victims into visiting malicious web pages leading to attacks like password theft or drive-by download. While recent reports indicate a surge of clickbait PDFs, prior works have largely neglected this new threat, considering PDFs only as accessories of email phishing campaigns.

Misinformation is now a major problem due to its poten-tial high risks to our core democratic and societal values and orders. Out-of-context misinformation is one of the easiest and effective ways used by adversaries to spread vi-ral false stories. In this threat, a real image is re-purposed to support other narratives by misrepresenting its context and/or elements. The internet is being used as the go-to way to verify information using different sources and modali-ties. Our goal is an inspectable method that automates this time-consuming and reasoning-intensive process by fact-checking the image-caption pairing using Web evidence. To integrate evidence and cues from both modalities, we intro-duce the concept of ‘multi-modal cycle-consistency check’ starting from the image/caption, we gather tex-tual/visual evidence, which will be compared against the other paired caption/image, respectively. Moreover, we propose a novel architecture, Consistency-Checking Network (CCN), that mimics the layered human reasoning across the same and different modalities: the caption vs. textual evidence, the image vs. visual evidence, and the image vs. caption. Our work offers the first step and bench-mark for open-domain, content-based, multi-modal fact-checking, and significantly outperforms previous baselines that did not leverage external evidence11For code, checkpoints, and dataset, check: https://s-abdelnabi.github.io/OoC-multi-modal-fc/.

Photorealistic image generation has reached a new level of quality due to the breakthroughs of generative adversarial networks (GANs). Yet, the dark side of such deepfakes, the malicious use of generated media, raises concerns about visual misinformation. While existing research work on deepfake detection demonstrates high accuracy, it is subject to advances in generation techniques and adversarial iterations on detection countermeasure techniques. Thus, we seek a proactive and sustainable solution on deepfake detection, that is agnostic to the evolution of generative models, by introducing artificial fingerprints into the models.Our approach is simple and effective. We first embed artificial fingerprints into training data, then validate a surprising discovery on the transferability of such fingerprints from training data to generative models, which in turn appears in the generated deepfakes. Experiments show that our fingerprinting solution (1) holds for a variety of cutting-edge generative models, (2) leads to a negligible side effect on generation quality, (3) stays robust against image-level and model-level perturbations, (4) stays hard to be detected by adversaries, and (5) converts deepfake detection and attribution into trivial tasks and outperforms the recent state-of-the-art baselines. Our solution closes the responsibility loop between publishing pre-trained generative model inventions and their possible misuses, which makes it independent of the current arms race.

Recent advances in natural language generation have introduced powerful language models with high-quality output text. However, this raises concerns about the potential misuse of such models for malicious purposes. In this paper, we study natural language watermarking as a defense to help better mark and trace the provenance of text. We introduce the Adversarial Watermarking Transformer (AWT) with a jointly trained encoder-decoder and adversarial training that, given an input text and a binary message, generates an output text that is unobtrusively encoded with the given message. We further study different training and inference strategies to achieve minimal changes to the semantics and correctness of the input text.AWT is the first end-to-end model to hide data in text by automatically learning -without ground truth- word substitutions along with their locations in order to encode the message. We empirically show that our model is effective in largely preserving text utility and decoding the watermark while hiding its presence against adversaries. Additionally, we demonstrate that our method is robust against a range of attacks.

Phishing websites are still a major threat in today's Internet ecosystem. Despite numerous previous efforts, similarity-based detection methods do not offer sufficient protection for the trusted websites, in particular against unseen phishing pages. This paper contributes VisualPhishNet, a new similarity-based phishing detection framework, based on a triplet Convolutional Neural Network (CNN). VisualPhishNet learns profiles for websites in order to detect phishing websites by a similarity metric that can generalize to pages with new visual appearances. We furthermore present VisualPhish, the largest dataset to date that facilitates visual phishing detection in an ecologically valid manner. We show that our method outperforms previous visual similarity phishing detection approaches by a large margin while being robust against a range of evasion attacks.

Despite significant advances in using Steady-State Visually Evoked Potentials (SSVEP) for on-screen target discrimination, existing methods either require intrusive, low-frequency visual stimulation or only support a small number of targets. We propose SSVEPNet: a convolutional long short-term memory (LSTM) recurrent neural network for high-frequency stimulation (≥30Hz) using a large number of visual targets. We evaluate our method for discriminating between 43 targets on an extended alphanumeric virtual keyboard and compare three different frequency assignment strategies. Our experimental results show that SSVEPNet significantly outperforms state-of-the-art correlation-based methods and convolutional neural networks. As such, our work opens up an exciting new direction of research towards a new class of unobtrusive and highly expressive SSVEP-based interfaces for text entry and beyond.