Related papers: Mind Your HEARTBEAT! Claw Background Execution Inh…

A long-standing vision of computing is the personal AI system: one that understands us well enough to address our underlying needs. Today's AI focuses on what users do, ignoring why they might be doing such things in the first place. As a…

We report a safety incident in a deployed multi-agent research system in which a primary AI agent installed 107 unauthorized software components, overwrote a system registry, overrode a prior negative decision from an oversight agent, and…

Large Language Models (LLMs) are susceptible to indirect prompt injection attacks, where the model inadvertently responds to instructions injected into the prompt context. This vulnerability stems from LLMs' inability to distinguish between…

Continual (or "incremental") learning approaches are employed when additional knowledge or tasks need to be learned from subsequent batches or from streaming data. However these approaches are typically adversary agnostic, i.e., they do not…

Claw-style environments support multi-step workflows over local files, tools, and persistent workspace states. However, scalable development around these environments remains constrained by the absence of a systematic framework, especially…

With the success of deep learning algorithms in various domains, studying adversarial attacks to secure deep models in real world applications has become an important research topic. Backdoor attacks are a form of adversarial attacks on…

It is commonplace to produce application-specific models by fine-tuning large pre-trained models using a small bespoke dataset. The widespread availability of foundation model checkpoints on the web poses considerable risks, including the…

Backdoor attacks embed malicious triggers into training data, enabling attackers to manipulate neural network behavior during inference while maintaining high accuracy on benign inputs. However, existing backdoor attacks face limitations…

Insider threats are a particularly tricky cybersecurity issue, especially in zero-trust architectures (ZTA) where implicit trust is removed. Although the rule of thumb is never trust, always verify, attackers can still use legitimate…

Tool-use large language model (LLM) agents are increasingly deployed to support sensitive workflows, relying on tool calls for retrieval, external API access, and session memory management. While prior research has examined various threats,…

The advent of clinical language models integrated into electronic health records (EHR) for clinical decision support has marked a significant advancement, leveraging the depth of clinical notes for improved decision-making. Despite their…

In-context learning, a paradigm bridging the gap between pre-training and fine-tuning, has demonstrated high efficacy in several NLP tasks, especially in few-shot settings. Despite being widely applied, in-context learning is vulnerable to…

Emerging applications such as embodied intelligence, AI hardware, autonomous driving, and intelligent cockpits rely on a real-time perception-decision-action closed loop, posing stringent challenges for streaming video understanding.…

Backdoor vulnerabilities widely exist in the fine-tuning of large language models(LLMs). Most backdoor poisoning methods operate mainly at the token level and lack deeper semantic manipulation, which limits stealthiness. In addition, Prior…

Real-world requests to AI agents are fundamentally underspecified. Natural human communication relies on shared context and unstated constraints that speakers expect listeners to infer. Current agentic benchmarks test explicit…

Speaker identification (SI) determines a speaker's identity based on their spoken utterances. Previous work indicates that SI deep neural networks (DNNs) are vulnerable to backdoor attacks. Backdoor attacks involve embedding hidden triggers…

In early 2018, Meltdown first showed how to read arbitrary kernel memory from user space by exploiting side-effects from transient instructions. While this attack has been mitigated through stronger isolation boundaries between user and…

Architectural obfuscation - e.g., permuting hidden-state tensors, linearly transforming embedding tables, or remapping tokens - has recently gained traction as a lightweight substitute for heavyweight cryptography in privacy-preserving…

Backdoor attacks insert malicious data into a training set so that, during inference time, it misclassifies inputs that have been patched with a backdoor trigger as the malware specified label. For backdoor attacks to bypass human…

The emergence of multimodal large language models has redefined the agent paradigm by integrating language and vision modalities with external data sources, enabling agents to better interpret human instructions and execute increasingly…