arXiv subjects

Vitaly Shmatikov

Publications and source records attributed to Vitaly Shmatikov.

At least 19 recordsLinked to original sources

arXiv2026

Deep-Research Agents Can Be Poisoned via User-Generated Content

Deep-research agents are an alternative to conventional Web search. They use multi-agent pipelines to issue multiple Web searches related to user queries, retrieve relevant content from the answers, and generate detailed, evidence-based reports. We show that for many common search topics (including financial, medical, and product recommendations), agent-generated reports are consistently based on the same user-generated content (UGC) pages from platforms such as Reddit and Wikipedia. This retrieval overlap, combined with lax moderation and access controls for UGC, is an opportunity for subversion: an attacker who appends a short text to a single, frequently-retrieved page can cause agents to cite this text and promote attacker-chosen entities across many user questions. We evaluate this attack on representative deep-research systems across multiple topics, such as investment advice, antivirus software, restaurant recommendations, etc. Poisoning a single URL with as few as 13 words can be sufficient for the attacker's content to be retrieved in 57-76% of the agent executions for a given topic and cited in 38-51% of the generated reports. A more aggressive attack (poisoning an entire subreddit) achieves 30-53% citation rates even when the poison is only 0.5-4% of the retrieved content. The attacker does not need to know the phrasing of the user's question, nor the specific Web queries generated by the agent, nor the agent's internal retrieval and generation mechanisms. We then study defenses at different stages of the pipeline, including source-level filtering and output-based detection. Dropping UGC from retrieved content blocks the attack but degrades the quality of generated reports. We show that lightweight anomaly detection on inputs and outputs does not reliably identify poisoned content, nor the results of poisoning.

cs.CR↗

arXiv2026

Speculative Probing: LLM Monitoring at Speculative-Decoding Cost

Real-time classification during language model inference is valuable for safety filtering, behavioral analysis, and model monitoring, but current approaches force a trade-off between accuracy and efficiency. Hidden-state probes are fast but limited: they are either not context-aware: operating on a single vector and cannot model interactions across positions; or they are very costly: having dedicated classifier models (Llama Guard, Qwen Guard, LLM-as-judge) or performing computation on hidden states for all tokens and then pooling the results (MultiMax). This shows an intrinsic trade-off between efficiency and accuracy. However, we find that the speculative-decoding module in recent LLMs can be repurposed for efficient high-quality classification. By appending a trained soft prompt at the end of the target sequence, we can repurpose the speculative-decoding module into a sequence classifier. At inference time in a speculative-decoding pipeline, the KV cache is already in GPU memory, so classification adds negligible overhead. We evaluate on four classification tasks across four models (Qwen3.5-4B, 9B, 27B, MiniCPM4.1-8B). Our small probes consistently outperform zero-shot GPT-5.4-mini and, on multilingual prompt safety, match or beat specialized 8B safety classifiers (Qwen3Guard-Gen-8B, Llama-Guard-3-8B) without running a full LLM.

cs.AI↗

arXiv2026

Agent Meltdowns: The Road to Hell Is Paved with Helpful Agents

Agents operating with computer and Web use inevitably encounter errors: inaccessible webpages, missing files, local and remote misconfigurations, etc. These errors do not thwart agents based on state-of-the-art models. They helpfully continue to look for ways to complete their tasks. We introduce, characterize, and measure a new type of agent failure we call \emph{accidental meltdown}: unsafe or harmful behavior in response to a benign environmental error, in the absence of any adversarial inputs. Because meltdowns are not captured by the existing reliability or safety benchmarks, we develop a taxonomy of meltdown behaviors. We then implement an agent-agnostic infrastructure for injecting simulated local and remote errors into the rollout environment and use it to systematically evaluate agent systems powered by GPT, Grok, and Gemini. Our evaluation demonstrates that meltdowns (e.g., conducting unauthorized reconnaissance or subverting access control) of varying severity and success occur in 64.7\% of agent rollouts that encounter simulated errors, spanning all combinations of agent system, backing model, and error type. In over half of these meltdowns, unsafe behaviors are not reported to the user. Comparing behaviors of the same agents with and without errors, we find that exploration in response to errors is correlated with unsafe and harmful behavior.

cs.CL↗

arXiv2026

Adversarial Hubness in Multi-Modal Retrieval

Hubness is a phenomenon in high-dimensional vector spaces where a point from the natural distribution is unusually close to many other points. This is a well-known problem in information retrieval that causes some items to accidentally (and incorrectly) appear relevant to many queries. In this paper, we investigate how attackers can exploit hubness to turn any image or audio input in a multi-modal retrieval system into an adversarial hub. Adversarial hubs can be used to inject universal adversarial content (e.g., spam) that will be retrieved in response to thousands of different queries, and also for targeted attacks on queries related to specific, attacker-chosen concepts. We present a method for creating adversarial hubs and evaluate the resulting hubs on benchmark multi-modal retrieval datasets and an image-to-image retrieval system implemented by Pinecone, a popular vector database. For example, in text-caption-to-image retrieval, a single adversarial hub, generated using 100 random queries, is retrieved as the top-1 most relevant image for more than 21,000 out of 25,000 test queries (by contrast, the most common natural hub is the top-1 response to only 102 queries), demonstrating the strong generalization capabilities of adversarial hubs. We also investigate whether techniques for mitigating natural hubness can also mitigate adversarial hubs, and show that they are not effective against hubs that target queries related to specific concepts.

cs.CR↗

arXiv2026

How to Steal Reasoning Without Reasoning Traces

Many large language models (LLMs) use reasoning to generate responses but do not reveal their full reasoning traces (a.k.a. chains of thought), instead outputting only final answers and brief reasoning summaries. To demonstrate that hiding reasoning traces does not prevent users from "stealing" a model's reasoning capabilities, we introduce trace inversion models that, given only the inputs, answers, and (optionally) reasoning summaries exposed by a target model, generate detailed, synthetic reasoning traces. We show that (1) traces synthesized by trace inversion have high overlap with the ground-truth reasoning traces (when available), and (2) fine-tuning student models on inverted traces substantially improves their reasoning and enables distillation from proprietary, black-box LLMs.

cs.CR↗

arXiv2026

Breaking and Fixing Defenses Against Control-Flow Hijacking in Multi-Agent Systems

Control-flow hijacking attacks manipulate orchestration mechanisms in multi-agent systems into performing unsafe actions that compromise the system and exfiltrate sensitive information. Recently proposed defenses, such as LlamaFirewall, rely on alignment checks of inter-agent communications to ensure that all agent invocations are "related to" and "likely to further" the original objective. We start by demonstrating control-flow hijacking attacks that evade these defenses even if alignment checks are performed by advanced LLMs. We argue that the safety and functionality objectives of multi-agent systems fundamentally conflict with each other. This conflict is exacerbated by the brittle definitions of "alignment" and the checkers' incomplete visibility into the execution context. We then propose, implement, and evaluate ControlValve, a new defense inspired by the principles of control-flow integrity and least privilege. ControlValve (1) generates permitted control-flow graphs for multi-agent systems, and (2) enforces that all executions comply with these graphs, along with contextual rules (generated in a zero-shot manner) for each agent invocation.

cs.LG↗

arXiv2026

Learning to Detect Language Model Training Data via Active Reconstruction

Detecting LLM training data is generally framed as a membership inference attack (MIA) problem. However, conventional MIAs operate passively on fixed model weights, using log-likelihoods or text generations. In this work, we introduce \textbf{Active Data Reconstruction Attack} (ADRA), a family of MIA that actively induces a model to reconstruct a given text through training. We hypothesize that training data are \textit{more reconstructible} than non-members, and the difference in their reconstructibility can be exploited for membership inference. Motivated by findings that reinforcement learning (RL) sharpens behaviors already encoded in weights, we leverage on-policy RL to actively elicit data reconstruction by finetuning a policy initialized from the target model. To effectively use RL for MIA, we design reconstruction metrics and contrastive rewards. The resulting algorithms, \textsc{ADRA} and its adaptive variant \textsc{ADRA+}, improve both reconstruction and detection given a pool of candidate data. Experiments show that our methods consistently outperform existing MIAs in detecting pre-training, post-training, and distillation data, with an average improvement of 10.7\% over the previous runner-up. In particular, \MethodPlus~improves over Min-K\%++ by 18.8\% on BookMIA for pre-training detection and by 7.6\% on AIME for post-training detection.

cs.LG↗

arXiv2026

Harnessing the Universal Geometry of Embeddings

We introduce the first method for translating text embeddings from one vector space to another without any paired data, encoders, or predefined sets of matches. Our unsupervised approach translates any embedding to and from a universal latent representation (i.e., a universal semantic structure conjectured by the Platonic Representation Hypothesis). Our translations achieve high cosine similarity across model pairs with different architectures, parameter counts, and training datasets. The ability to translate unknown embeddings into a different space while preserving their geometry has serious implications for the security of vector databases. An adversary with access only to embedding vectors can extract sensitive information about the underlying documents, sufficient for classification and attribute inference.

cs.LG↗

arXiv2025

Machine Unlearning Doesn't Do What You Think: Lessons for Generative AI Policy and Research

"Machine unlearning" is a popular proposed solution for mitigating the existence of content in an AI model that is problematic for legal or moral reasons, including privacy, copyright, safety, and more. For example, unlearning is often invoked as a solution for removing the effects of specific information from a generative-AI model's parameters, e.g., a particular individual's personal data or the inclusion of copyrighted content in the model's training data. Unlearning is also proposed as a way to prevent a model from generating targeted types of information in its outputs, e.g., generations that closely resemble a particular individual's data or reflect the concept of "Spiderman." Both of these goals--the targeted removal of information from a model and the targeted suppression of information from a model's outputs--present various technical and substantive challenges. We provide a framework for ML researchers and policymakers to think rigorously about these challenges, identifying several mismatches between the goals of unlearning and feasible implementations. These mismatches explain why unlearning is not a general-purpose solution for circumscribing generative-AI model behavior in service of broader positive impact.

cs.LG↗

arXiv2025

MillStone: How Open-Minded Are LLMs?

Large language models equipped with Web search, information retrieval tools, and other agentic capabilities are beginning to supplant traditional search engines. As users start to rely on LLMs for information on many topics, including controversial and debatable issues, it is important to understand how the stances and opinions expressed in LLM outputs are influenced by the documents they use as their information sources. In this paper, we present MillStone, the first benchmark that aims to systematically measure the effect of external arguments on the stances that LLMs take on controversial issues (not all of them political). We apply MillStone to nine leading LLMs and measure how ``open-minded'' they are to arguments supporting opposite sides of these issues, whether different LLMs agree with each other, which arguments LLMs find most persuasive, and whether these arguments are the same for different LLMs. In general, we find that LLMs are open-minded on most issues. An authoritative source of information can easily sway an LLM's stance, highlighting the importance of source selection and the risk that LLM-based information retrieval and search systems can be manipulated.

cs.LG↗

arXiv2025

Multi-Agent Systems Execute Arbitrary Malicious Code

Multi-agent systems coordinate LLM-based agents to perform tasks on users' behalf. In real-world applications, multi-agent systems will inevitably interact with untrusted inputs, such as malicious Web content, files, email attachments, and more. Using several recently proposed multi-agent frameworks as concrete examples, we demonstrate that adversarial content can hijack control and communication within the system to invoke unsafe agents and functionalities. This results in a complete security breach, up to execution of arbitrary malicious code on the user's device or exfiltration of sensitive data from the user's containerized environment. For example, when agents are instantiated with GPT-4o, Web-based attacks successfully cause the multi-agent system execute arbitrary malicious code in 58-90\% of trials (depending on the orchestrator). In some model-orchestrator configurations, the attack success rate is 100\%. We also demonstrate that these attacks succeed even if individual agents are not susceptible to direct or indirect prompt injection, and even if they refuse to perform harmful actions. We hope that these results will motivate development of trust and security models for multi-agent systems before they are widely deployed.

cs.CR↗

arXiv2025

Adversarial Illusions in Multi-Modal Embeddings

Multi-modal embeddings encode texts, images, thermal images, sounds, and videos into a single embedding space, aligning representations across different modalities (e.g., associate an image of a dog with a barking sound). In this paper, we show that multi-modal embeddings can be vulnerable to an attack we call "adversarial illusions." Given an image or a sound, an adversary can perturb it to make its embedding close to an arbitrary, adversary-chosen input in another modality. These attacks are cross-modal and targeted: the adversary can align any image or sound with any target of his choice. Adversarial illusions exploit proximity in the embedding space and are thus agnostic to downstream tasks and modalities, enabling a wholesale compromise of current and future tasks, as well as modalities not available to the adversary. Using ImageBind and AudioCLIP embeddings, we demonstrate how adversarially aligned inputs, generated without knowledge of specific downstream tasks, mislead image generation, text generation, zero-shot classification, and audio retrieval. We investigate transferability of illusions across different embeddings and develop a black-box version of our method that we use to demonstrate the first adversarial alignment attack on Amazon's commercial, proprietary Titan embedding. Finally, we analyze countermeasures and evasion attacks.

cs.CR↗

arXiv2025

Approximating Language Model Training Data from Weights

Modern language models often have open weights but closed training data. We formalize the problem of data approximation from model weights and propose several baselines and metrics. We develop a gradient-based approach that selects the highest-matching data from a large public text corpus and show its effectiveness at recovering useful data given only weights of the original and finetuned models. Even when none of the true training data is known, our method is able to locate a small subset of public Web documents can be used to train a model to close to the original model performance given models trained for both classification and supervised-finetuning. On the AG News classification task, our method improves performance from 65% (using randomly selected data) to 80%, approaching the expert benchmark of 88%. When applied to a model trained with SFT on MSMARCO web documents, our method reduces perplexity from 3.3 to 2.3, compared to an expert LLAMA model's perplexity of 2.0.

cs.CL↗

arXiv2025

Self-interpreting Adversarial Images

We introduce a new type of indirect, cross-modal injection attacks against visual language models that enable creation of self-interpreting images. These images contain hidden "meta-instructions" that control how models answer users' questions about the image and steer models' outputs to express an adversary-chosen style, sentiment, or point of view. Self-interpreting images act as soft prompts, conditioning the model to satisfy the adversary's (meta-)objective while still producing answers based on the image's visual content. Meta-instructions are thus a stronger form of prompt injection. Adversarial images look natural and the model's answers are coherent and plausible, yet they also follow the adversary-chosen interpretation, e.g., political spin, or even objectives that are not achievable with explicit text instructions. We evaluate the efficacy of self-interpreting images for a variety of models, interpretations, and user prompts. We describe how these attacks could cause harm by enabling creation of self-interpreting content that carries spam, misinformation, or spin. Finally, we discuss defenses.

cs.CR↗

arXiv2025

Universal Zero-shot Embedding Inversion

Embedding inversion, i.e., reconstructing text given its embedding and black-box access to the embedding encoder, is a fundamental problem in both NLP and security. From the NLP perspective, it helps determine how much semantic information about the input is retained in the embedding. From the security perspective, it measures how much information is leaked by vector databases and embedding-based retrieval systems. State-of-the-art methods for embedding inversion, such as vec2text, have high accuracy but require (a) training a separate model for each embedding, and (b) a large number of queries to the corresponding encoder. We design, implement, and evaluate ZSInvert, a zero-shot inversion method based on the recently proposed adversarial decoding technique. ZSInvert is fast, query-efficient, and can be used for any text embedding without training an embedding-specific inversion model. We measure the effectiveness of ZSInvert on several embeddings and demonstrate that it recovers key semantic information about the corresponding texts.

cs.CL↗

arXiv2025

Machine Against the RAG: Jamming Retrieval-Augmented Generation with Blocker Documents

Retrieval-augmented generation (RAG) systems respond to queries by retrieving relevant documents from a knowledge database and applying an LLM to the retrieved documents. We demonstrate that RAG systems that operate on databases with untrusted content are vulnerable to denial-of-service attacks we call jamming. An adversary can add a single ``blocker'' document to the database that will be retrieved in response to a specific query and result in the RAG system not answering this query, ostensibly because it lacks relevant information or because the answer is unsafe. We describe and measure the efficacy of several methods for generating blocker documents, including a new method based on black-box optimization. Our method (1) does not rely on instruction injection, (2) does not require the adversary to know the embedding or LLM used by the target RAG system, and (3) does not employ an auxiliary LLM. We evaluate jamming attacks on several embeddings and LLMs and demonstrate that the existing safety metrics for LLMs do not capture their vulnerability to jamming. We then discuss defenses against blocker documents.

cs.CR↗

arXiv2025

Adversarial Decoding: Generating Readable Documents for Adversarial Objectives

We design, implement, and evaluate adversarial decoding, a new, generic text generation technique that produces readable documents for different adversarial objectives. Prior methods either produce easily detectable gibberish, or cannot handle objectives that include embedding similarity. In particular, they only work for direct attacks (such as jailbreaking) and cannot produce adversarial text for realistic indirect injection, e.g., documents that (1) are retrieved in RAG systems in response to broad classes of queries, and also (2) adversarially influence subsequent generation. We also show that fluency (low perplexity) is not sufficient to evade filtering. We measure the effectiveness of adversarial decoding for different objectives, including RAG poisoning, jailbreaking, and evasion of defensive filters, and demonstrate that it outperforms existing methods while producing readable adversarial documents.

cs.CL↗

arXiv2025

Rerouting LLM Routers

LLM routers aim to balance quality and cost of generation by classifying queries and routing them to a cheaper or more expensive LLM depending on their complexity. Routers represent one type of what we call LLM control planes: systems that orchestrate use of one or more LLMs. In this paper, we investigate routers' adversarial robustness. We first define LLM control plane integrity, i.e., robustness of LLM orchestration to adversarial inputs, as a distinct problem in AI safety. Next, we demonstrate that an adversary can generate query-independent token sequences we call ``confounder gadgets'' that, when added to any query, cause LLM routers to send the query to a strong LLM. Our quantitative evaluation shows that this attack is successful both in white-box and black-box settings against a variety of open-source and commercial routers, and that confounding queries do not affect the quality of LLM responses. Finally, we demonstrate that gadgets can be effective while maintaining low perplexity, thus perplexity-based filtering is not an effective defense. We finish by investigating alternative defenses.

cs.CR↗