When AI Turns Against Itself: The Rise of Proxy Attacks
New research reveals a concerning vulnerability in AI systems where agents can be exploited to circumvent safety protocols and deliver malicious instructions.
Science
Stuck in a Rut: Why AI Agents Struggle to Explore
![Across all evaluated tasks, explore-exploit baselines consistently surpassed the performance of language models when operating under a query budget of [latex]N=48[/latex], demonstrating robustness to variations in parameter settings.](https://arxiv.org/html/2601.22345v1/x72.png)
A new evaluation benchmark reveals that current language models often fail to adequately explore interactive environments, leading to suboptimal decisions and a lack of adaptability.
Beyond Rules: When AI Learns to Design
![Generative ontologies transcend descriptive vocabularies by establishing constraints that enable large language models to function as active grammars for design creation, ensuring validity through a formalized system-a principle akin to establishing that [latex] \forall x \in V : \text{ontology}(x) \implies \text{validity}(x) [/latex], where <i>V</i> represents the vocabulary and validity is guaranteed by the ontological framework.](https://arxiv.org/html/2602.05636v1/figures/fig-defining-gen-ontology.png)
A new framework merges the power of large language models with structured knowledge to unlock creative design possibilities.
How Language Models Learn to Predict: A Hidden Geometric Order
New research reveals that the deepest layers of large language models organize information geometrically, and this structure directly powers their predictive abilities.
Mapping the Way Forward: AI-Powered Terrain Perception for Humanoid Robots
Researchers have developed a new deep learning framework that fuses data from lidar and depth sensors to create detailed terrain maps, enabling more stable and reliable locomotion for humanoid robots.
Who Decides What’s Real Online?
As social media bot detection becomes increasingly sophisticated, critical ethical questions about fairness, accountability, and transparency demand urgent attention.
Beyond Labels: Scaling Linguistic Insight with AI Agents
![The LinguistAgent architecture establishes a framework for reasoning about language through the formalization of linguistic structures, enabling the agent to decompose complex sentences into their constituent parts and derive meaning based on underlying grammatical relationships - a process fundamentally rooted in the principles of compositional semantics, akin to evaluating [latex] f(g(x)) [/latex] where <i>f</i> represents semantic interpretation and <i>g</i> syntactic parsing.](https://arxiv.org/html/2602.05493v1/x1.png)
A new platform leverages the power of large language models and multi-agent systems to automate complex linguistic tasks, offering a transparent and reproducible approach to annotation.
Two Hands Are Better Than One: Mastering Robotic Dexterity with Multimodal AI
Researchers have developed a new AI framework that enables robots to perform complex, two-handed tasks with greater precision and adaptability.
Seeing is Reasoning: An Agent That Learns to Look
Researchers have developed a new system, Weaver, that learns to actively gather visual evidence from videos to improve its reasoning abilities.
Drone Autonomy Takes Flight with AI Brains
Researchers are exploring how large language models can give drones the reasoning skills needed to navigate complex indoor environments without pre-mapping.