Related papers: Attacker Control and Impact for Confidentiality an…

Adversarial attacking aims to fool deep neural networks with adversarial examples. In the field of natural language processing, various textual adversarial attack models have been proposed, varying in the accessibility to the victim model.…

The incremental diffusion of machine learning algorithms in supporting cybersecurity is creating novel defensive opportunities but also new types of risks. Multiple researches have shown that machine learning methods are vulnerable to…

This work aims to solve a practical problem, i.e., how to quantify the risk brought upon a system by different attackers. The answer is useful for optimising resource allocation for system defence. Given a set of safety requirements, we…

Safe multi-agent coordination in uncertain environments can benefit from learning constraints from other agents. Implicitly communicating safety constraints through actions is a promising approach, allowing agents to coordinate and maintain…

Large language models (LLMs) are increasingly deployed in multi-agent systems where agents communicate in natural language to solve tasks jointly. A key capability in such systems is consensus formation, where agents iteratively exchange…

Access control encryption (ACE) enforces, through a sanitizer as the mediator, that only legitimate sender-receiver pairs can communicate, without the sanitizer knowing the communication metadata, including its sender and recipient…

Pretrained language models have significantly advanced performance across various natural language processing tasks. However, adversarial attacks continue to pose a critical challenge to systems built using these models, as they can be…

Non-interference is a semantic program property that assigns confidentiality levels to data objects and prevents illicit information flows from occurring from high to low security levels. In this paper, we present a novel security model for…

We study secure and undetectable communication in a world where governments can read all encrypted communications of citizens. We consider a world where the only permitted communication method is via a government-mandated encryption scheme,…

Todays industrial control systems consist of tightly coupled components allowing adversaries to exploit security attack surfaces from the information technology side, and, thus, also get access to automation devices residing at the…

This paper investigates a flow- and path-sensitive static information flow analysis. Compared with security type systems with fixed labels, it has been shown that flow-sensitive type systems accept more secure programs. We show that an…

Information security is facing increasingly severe challenges, and traditional protection means are difficult to cope with complex and changing threats. In recent years, as an emerging intelligent technology, large language models (LLMs)…

The deployment of control systems with network-connected components has made feedback control systems vulnerable to attacks over the network. This paper considers the problem of intrusion detection and mitigation in supervisory control…

Semantic control entails steering LM generations towards satisfying subtle non-lexical constraints, e.g., toxicity, sentiment, or politeness, attributes that can be captured by a sequence-level verifier. It can thus be viewed as sampling…

The Model Context Protocol (MCP) replaces static, developer-controlled API integrations with more dynamic, user-driven agent systems, which also introduces new security risks. As MCP adoption grows across community servers and major…

In this paper, we consider an adversarial scenario where one agent seeks to achieve an objective and its adversary seeks to learn the agent's intentions and prevent the agent from achieving its objective. The agent has an incentive to try…

The growing dependence on machine learning in real-world applications emphasizes the importance of understanding and ensuring its safety. Backdoor attacks pose a significant security risk due to their stealthy nature and potentially serious…

Written language contains stylistic cues that can be exploited to automatically infer a variety of potentially sensitive author information. Adversarial stylometry intends to attack such models by rewriting an author's text. Our research…

Dependence on information, including for some of the world's largest organisations such as governments and multi-national corporations, has grown rapidly in recent years. However, reports of information security breaches and their…

Backdoor attack is a severe threat to the trustworthiness of DNN-based language models. In this paper, we first extend the definition of memorization of language models from sample-wise to more fine-grained sentence element-wise (e.g.,…