New research on AI and security, selected for relevance to language models, AI agents and cybersecurity. Summaries are the authors’ or publisher’s own words, linked to the original.
Peer-reviewed
Effective enzyme engineering through modeling kinetic parameter changes upon mutations with geometric deep learning
A gait foundation model for prediction of multisystem health phenotypes from 3D skeletal motion
Towards ecologically meaningful foundation models
‘Mathematics is more than a mechanical process’ — why I think the field will survive OpenAI’s disruption
Exploring the sustainable scaling of AI dilemma: a projective study of corporations’ AI environmental impacts
Biological foundation models
New preprints (not yet peer-reviewed)
SAVeR$^2$: Reasoning-based Safety Alignment for Large Reasoning Models via Verifiable Rewards
Large Reasoning Models (LRMs) have achieved impressive performance via explicit Chain-of-Thought (CoT) reasoning, yet this process introduces critical safety risks. We formalize LRM safety objectives as requiring non-malicious reasoning and answers, safety consistency, and explicit intent identification.
ARMs: Adaptive Red-Teaming Agent against Multimodal Models with Plug-and-Play Attacks
As vision-language models (VLMs) gain prominence, their multimodal interfaces also introduce new safety vulnerabilities, making the safety evaluation challenging and critical. Existing red-teaming efforts are either restricted to a narrow set of adversarial patterns or depend heavily on manual engineering, lacking scalable exploration of emerging real-world VLM vulnerabilities.
TSQAgent: Rating Time Series Data Quality via Dedicated Agentic Reasoning
Assessing the quality of time series (TS) data is fundamental yet inherently challenging due to the multifaceted nature of quality dimensions. Recently, large language models (LLMs) have emerged as a promising paradigm for TS quality assessment via pairwise comparison and per-dimension evaluation.
SEABench: Benchmarking Endogenous Misalignment In Self-Evolving Agents
Self-evolving LLM agents have gained prominence for their ability to improve after deployment by modifying their harness, including their controller instructions, memory management protocols, and reusable tools and skills, in response to user and environment feedback. However, locally useful updates may persist into later tasks where they produce unsafe behavior, even without direct adversarial influence.
CST-Bench: Benchmarking Counterfactual Spatial Transition Reasoning with Causal Transition On-Policy Self-Distillation
Spatial reasoning is a fundamental capability of multimodal large language models (MLLMs), yet existing benchmarks mainly focus on static scene understanding. When objects are moved or swapped, models must infer not only spatial relations but also how interventions change the scene state.
MCPHallu: Benchmarking Reasoning, Execution, and Memory Hallucinations in MCP Agents
Large language model agents are increasingly deployed through the Model Context Protocol (MCP), which allows them to discover and invoke tools across multiple servers at runtime. Existing MCP benchmarks primarily measure task completion; where they diagnose failures, they do so after execution rather than building specific reasoning, execution, and memory failure risks into task construction.
MEvince: Knowledge-Constrained LLM Reasoning for Technique Identification from Malware Reports
Mapping behavioral intelligence in malware analysis reports to ATT&CK techniques supports threat analysis and defense, but reliable identification remains challenging. Supporting evidence is often scattered across long, heterogeneous reports, while closely related techniques share behavioral cues and differ only in subtle but decisive details.
Systematic Evaluation of Adversarial Jailbreaking Attacks and Defenses Across Seven Large Language Models
Large language models (LLMs) deployed in safety-critical settings must resist adversarial inputs designed to elicit harmful outputs. We present a systematic empirical evaluation of five adversarial jailbreaking attack families-direct requests, roleplay-based jailbreaking, hypothetical framing, instruction override, and obfuscation-across seven open-weight LLMs served locally: Llama-3-8B, Llama-3.1-8B, Llama-3.2-3B, Mistral-7B, Qwen2.5-7B, Gemma2-9B, and Phi-4-14B.