Alignment, interpretability, red-teaming, bias and privacy: the research on what these systems do when they misbehave.
arXiv:2607. 04034v1 Announce Type: cross Abstract: The language models that underpin agents have seen a rapid rise in performance on function calling benchmarks.
arXiv:2607. 03510v1 Announce Type: cross Abstract: Enterprise artificial intelligence is moving from experimentation into operational workflows.
arXiv:2602. 17750v2 Announce Type: replace-cross Abstract: A key problem of solid mechanics is the identification of the constitutive law of a material, that is, the relation between strain history and stress.
arXiv:2607. 05163v1 Announce Type: cross Abstract: AI systems may produce failures after deployment that pre-deployment safety assessments do not anticipate.
arXiv:2603. 06921v2 Announce Type: replace-cross Abstract: Safe navigation of autonomous robots remains one of the core challenges in the field, especially in dynamic and uncertain environments.
arXiv:2607. 02800v1 Announce Type: new Abstract: Induction heads are attention circuits believed to underlie in-context learning in transformers, yet a precise characterization of the estimators they implement remains elusive.
arXiv:2607. 04528v1 Announce Type: new Abstract: Software-agent benchmarks usually report whether an agent solves a task, but the agent reaches that outcome through a harness that controls what it sees, which actions it can take, which failures are repaired, which states are verified, and which evidence is logged.
arXiv:2607. 03177v1 Announce Type: cross Abstract: Traditional reinforcement learning (RL) for recovery in autonomous systems lacks causal understanding and generalizes poorly to novel failure scenarios.
arXiv:2601. 22652v2 Announce Type: replace-cross Abstract: Spectral gradient methods, such as the Muon optimizer, modify gradient updates by preserving directional information while discarding scale, and have shown strong empirical performance in deep learning.
arXiv:2607. 02964v1 Announce Type: cross Abstract: A central goal of mechanistic interpretability is to understand how neural networks work and what each individual component does.
arXiv:2607. 02932v1 Announce Type: cross Abstract: Privacy is an important challenge when users interact with AI chatbots, since users may share sensitive information, explicitly or implicitly, and AI chatbots can use this information for user profiling.
arXiv:2603. 14504v2 Announce Type: replace-cross Abstract: Optimizing the noise samples of diffusion and flow models is an increasingly popular approach to align these models to target rewards at inference time.
arXiv:2505. 23847v5 Announce Type: replace-cross Abstract: Large language models (LLMs) are rapidly evolving into autonomous agents that cooperate across organizational boundaries, enabling joint disaster response, supply-chain optimization, and other tasks that demand decentralized expertise without surrendering data ownership.
arXiv:2602. 24115v2 Announce Type: replace Abstract: Open RAN (O-RAN) exposes rich control and telemetry interfaces across the Non-RT RIC, Near-RT RIC, and distributed units, but also makes it harder to operate multi-tenant, multi-objective RANs in a safe and auditable manner.
arXiv:2607. 02903v1 Announce Type: cross Abstract: Explainability is central to building trustworthy AI, yet explanation interfaces can inadvertently provide adversaries with an expanded privacy-related attack surfaces.
arXiv:2508. 16181v2 Announce Type: replace-cross Abstract: Cross-organizational collaboration in Model-Based Systems Engineering (MBSE) faces many challenges in achieving semantic alignment across independently developed system models.
Recent advancements in Multimodal Large Language Models (MLLMs) have evolved from static perception to interleaved visual-language reasoning, often referred to as ``thinking with images''. A basic operation in this reasoning process is to zoom in on regions of interest (often represented with bounding boxes) to acquire finer visual details.
Large language model (LLM) agents are increasingly evaluated on their ability to use tools, plan multi-step tasks, coordinate with other agents, and operate over extended horizons. Reported benchmark gains often obscure recurring failure modes documented across otherwise unrelated evaluation efforts.
Large language models (LLMs) increasingly issue judgments read as binary verdicts, and a growing literature reports such judgments shifting under logically irrelevant changes of wording - among them an amplified yes-no bias on moral dilemmas, absent in humans. A single framing cannot say what such a shift is: in a yes/no question the word "no" is at once logical verdict, lexical token, and last-printed option.
Model-specific adversarial attacks have been extensively studied. We study a different failure mode: naturally occurring statistical signals in vision data that can behave like backdoor-like triggers without being maliciously inserted.