Alignment, interpretability, red-teaming, bias and privacy: the research on what these systems do when they misbehave.
OpenAI is enhancing monitoring, alignment, and security for frontier AI models. The company’s new safeguards are shaping how quickly these models are developed. This approach reflects a focus on responsible advancement of AI capabilities.
TRUSS is a framework that generates and verifies automated agent skills, ensuring they are both functionally effective and safe. It first checks functional claims against evidence and evaluates artifacts against nine safety properties, then tests admitted skills in a controlled environment to capture execution traces and identify failures. The approach achieves perfect precision and recall in vulnerability detection, significantly reduces attack success rates, and boosts task effectiveness and security rates in skill generation benchmarks.
Reflex-Guard is a lightweight, locally running guardrail for large language models that uses jailbreak-aware preprocessing, compact sentence‑transformer embeddings, and seven fast binary classifiers to filter unsafe prompts. It achieves 95.9% recall on harmful prompts with an end‑to‑end latency of 37.6 ms, far faster than existing solutions such as Llama Guard 2 (255 ms) and SafeDecoding (723 ms). The system can detect all GCG suffix attacks and Base64‑encoded prompts at the default threshold, and it attains a Reflex Efficiency Score up to 16.79, outperforming its competitors.
The paper introduces a spatio‑temporal traffic forecasting framework that fuses Graph Neural Networks with embeddings derived from a general‑purpose knowledge graph such as Wikidata. By creating semantic subgraphs around traffic sensors, the approach captures relationships like nearby points of interest, administrative hierarchies, and functional roles of locations, which are then integrated as additional adjacency matrices for the GNN. Experiments demonstrate that this external knowledge improves prediction accuracy beyond what street‑network data alone can achieve, while also offering a path toward better interpretability.
The paper introduces Fair-ASR, a new evaluation protocol for black-box jailbreak attacks that uses shared target-call budgets to provide a fair comparison across methods. Re‑evaluating 11 attacks under this protocol shows that rankings shift significantly with different budgets, and that simple perturbations and templates remain competitive. The authors also present ReCode, a budget‑efficient attack that combines desensitization rewriting with low‑cost primitives, achieving 85% ASR on GPT‑5 with only 7.19 attacker calls per request under a 20‑call budget.
arXiv:2608. 14590v1 Announce Type: new Abstract: LLM agents increasingly perform irreversible real-world actions, including database updates, API calls, file operations, and autonomous use of tools.
arXiv:2608. 15402v1 Announce Type: new Abstract: Generative model alignment has received broad interest, and significant progress has been made in supervised fine-tuning and inference-time computation.
arXiv:2608. 14820v1 Announce Type: cross Abstract: Handover (HO) management in vehicular networks requires fast and reliable decision-making under highly dynamic conditions.
arXiv:2608. 14825v1 Announce Type: cross Abstract: Frontier LLM agents increasingly transact on behalf of separate principals, often using natural language rather than structured APIs.
arXiv:2608. 14684v1 Announce Type: cross Abstract: LLM judges increasingly evaluate responses against fine-grained rubric checklists.
arXiv:2608. 15012v1 Announce Type: cross Abstract: The rapid advancement of large language models (LLMs) has created a growing asymmetry in cybersecurity, where attack accelerates toward autonomous execution while defense remains predominantly human-intensive.
arXiv:2608. 16287v1 Announce Type: new Abstract: Joint-embedding predictive world models plan by scoring predicted terminal embeddings against a goal embedding using a cost defined on the representation itself.
arXiv:2608. 16349v1 Announce Type: new Abstract: Large language model (LLM) agents may assist flight crews with complex decisions and task execution, but existing aviation evaluations centered on static knowledge do not support systematic testing of procedural execution and safety compliance in interactive environments.
arXiv:2608. 16465v1 Announce Type: new Abstract: Automated red-teaming has produced a growing collection of attack strategies, yet they typically remain scattered across prompts and workflows, making them difficult to systematically integrate, reuse, and improve at scale.
arXiv:2608. 16236v1 Announce Type: new Abstract: Collaborative inference (CI) splits a model between an edge device and a server, whereby the client computes an intermediate activation, transmits it, and the server completes the computation.
arXiv:2608. 16005v1 Announce Type: new Abstract: This paper studies the problem of molecule-text alignment, which aims to project molecules and their textual descriptions into a joint latent space for downstream tasks including molecule search and molecular property prediction.
arXiv:2608. 15996v1 Announce Type: new Abstract: We study second-order path-length regret in adversarial $K$-armed bandits against oblivious loss sequences.
arXiv:2605. 08974v2 Announce Type: replace-cross Abstract: While multimodal large language models (MLLMs) have advanced video understanding, they remain highly prone to hallucinations in dynamic scenes.
arXiv:2602. 21819v3 Announce Type: replace-cross Abstract: Reconstructing dynamic visual experiences from brain activity provides a compelling avenue for exploring the neural mechanisms of human visual perception.
arXiv:2602. 21219v2 Announce Type: replace-cross Abstract: Large Language Model (LLM) personalization holds great promise for tailoring responses by leveraging personal context and history.