AI
Offering Zero Data Retention for frontier models
OpenAI has announced that it will retain no user data from its frontier models, a move aimed at prioritizing data privacy. This means that users who a...
AI
AI
Replit expands access to software creation with GPT-5.6 Luna
Replit has introduced a new mode called 'Free Mode', which allows users to create software without incurring token costs. This is made possible by the...
AI
GxP-Agent: Process-DAG Topology for Reliable Clinical Trial Programming with LLM Agents
Researchers have developed GxP-Agent, a system to help with regulatory submissions in clinical trials. Current approaches using large language models ...
AI
Runtime Governance for Agentic AI: Action-Boundary Control with Trusted Provenance and Fail-Closed Execution
Researchers have developed a system called Aegis to govern the actions of agentic AI systems. These systems can request tool actions that modify files...
AI
The Price of Thinking: Reasoning Effort as a Model-Specific API Contract
Researchers have found that including a 'reasoning-effort term' in an AI model's contract can lead to a small increase in accuracy, but also a signifi...
AI
FedPref: Federated Preference Learning for Structured Radiology Report Extraction
Researchers have developed a new method for extracting structured information from radiology reports. The approach, called FedPref, allows institution...
AI
The Problem Is the Problem: Towards Scalable Mathematical Discovery
Researchers propose a new approach to AI-assisted mathematical discovery. Instead of selecting problems in advance, humans provide a research directio...
AI
SkillEffect: Checked Lowering for Memory-Bounded Agent Tools
Researchers have developed a system called SkillEffect to help language models manage memory usage when interacting with tools. The system uses a chec...
AI
Memory Is Communication: The Frontier Between Remembering and Signaling
Researchers propose a new framework to understand how agents allocate information between their own memory and communication with peers. They introduc...
AI
DiSCO: Defending text-to-image generation through distribution-guided contrastive prompt optimization
Researchers have proposed a new defense mechanism called DiSCO to prevent text-to-image generative models from producing harmful content. The system o...
AI
KernelArc: A Multi-Agent Framework for GPU Kernel Optimization
Researchers have developed KernelArc, a framework for optimizing GPU kernels across different workloads. The system uses multiple agents that run in p...
AI
A decodability criterion predicts when hidden-state selection beats majority voting in large language models
Researchers have proposed a method called CASE to improve the accuracy of large language models by selecting the best answer based on hidden-state sig...