HyperPersona introduces a hypergraph-based framework for text-based automatic personality prediction. The model explicitly captures language hierarchy (document, sentence, word) through hypergraph structure, then applies a transformer-based graph encoder to model multi-level dependencies. Achieves superior performance on Big Five personality dimensions.
Paper presents an AI agent with semantic and episodic memory systems inspired by cognitive science. Authors design and release "The Room" environment (OpenAI Gym-compatible) showing that combining both memory types outperforms single systems. Multi-agent collaboration improves performance.
AAMLA, a multimodal learning analytics framework, predicts student collaboration satisfaction in game-based educational environments. The CAMA module aligns modalities (gaze, action units, pose) via affinity matrices and contrastive learning, adaptively suppressing uninformative modalities. Tests on 50 middle school students in EcoJourneys show improvement over unimodal baselines.
PACC (Prosodic Adjustment with Conversational Context) converts read speech to natural conversational speech using deep neural networks and HiFi-GAN. The approach improves intonation, stress, and rhythm for real-time applications (virtual assistants, customer service). New benchmarks established on MOS and speech conversion tasks.
ML study on pre-test risk stratification for Chlamydia trachomatis infection. 93 urine samples analyzed using patient-reported data and urine biomarkers. Combined models reach AUC 0.72; urine biomarkers provide reliable predictive signal complementary to clinical data.
Scania scales ChatGPT Enterprise globally to boost workforce productivity with team-based onboarding and security guardrails. Use cases span quality improvement and innovation across manufacturing operations.
Optimus, a semi-supervised learning method for graphs, matches GCN performance with 5× fewer labels. On PathMNIST (2000 nodes, 9 classes), with 9 labels Optimus achieves 73.9% vs 60.6% for GCN. Live demo on Hugging Face Spaces.
OpenLumara is an open-source (GPL2) AI agent designed for local models, featuring a ~4k-token system prompt and modular architecture. Developed manually over months, it prioritizes token efficiency, built-in security, and user-friendly WebUI, positioning itself as lighter and more secure than OpenClaw or Hermes.
llama-bench benchmarking on AMD R9700 32GB with three Qwen3 models (8B, 14B, 32B in Q4_K_M quantization). Performance comparison across different hardware setups with detailed results published.
Spice, an open-source project, introduces an explicit decision layer above AI agents. It decouples decision-making (observation, options, justification, trade-offs) from execution, making agent behavior less of a black box. Compatible with Claude Code, Codex, and other existing agents.
White House executive order requires federal agencies (Pentagon, CISA) to strengthen cyber defense with AI tools within 30 days. AI developers can voluntarily submit models for security testing, but the order explicitly rules out mandatory approval. The genuinely voluntary nature of this cooperation remains unclear.
Linux ROCm now supports WSL2 with stable functionality, though bugs remain. Build instructions included for users.
BYD launches its Xuanji A3 chip for autonomous driving, reducing reliance on Nvidia. The Chinese automaker aims to control its value chain by developing proprietary autonomous driving solutions.
Unsloth Studio now fully supports training with MLX on Mac. The feature, previously marked as "coming soon", is now available in recent GitHub updates.
Comparative benchmark of vector search libraries (FAISS, Scann, Usearch) measuring speed, memory usage, and similarity accuracy. Tests across 500 to 1 million samples. Results and code published on GitHub.
Researcher launches red-teaming challenge on Llama 3.1 8B to stress-test SAFi, a runtime governance engine designed to enforce alignment of autonomous agents. 10 prompts to break a Socratic Tutor Agent (make it give direct answers or go off-topic from science/math). Open-source code available.
Tool and documentation to check OpenAI compatibility of open-source projects (vLLM, llama.cpp, etc.). Documents official and unofficial signatures, with extensions for other model types. Useful for integrating LLM endpoints into applications or building proxies/middleware.
Developer open-sources a background removal tool built on latest open source models, originally created for personal workflow. Tool now works as headless MCP service for agents. README generated with Gemini Flash.
Mid-year report on Colleague AI usage across 12 Washington State school districts (September–December 2025). Joint study by Colleague AI and AmplifyLearn.AI (University of Washington) analyzing teacher engagement with generative AI in K-12 classrooms.
Paper examining LLM integration in qualitative research. Identifies critical parameters (context window, temperature, top-p, prompt design, system cards) and highlights tensions between modern LLM opacity and qualitative epistemological principles (reflexivity, positionality, interpretive judgment).
OpenAI explains ChatGPT's privacy safeguards: reduced personal data in training, user control over conversation usage for model improvement, and protective mechanisms during learning processes.
Hugging Face publishes a guide for integrating Transformers.js into a Chrome extension. The library enables running transformer models directly in the browser without a backend server.
OpenAI and Gates Foundation run a workshop in Asia to help disaster response teams operationalize AI tools. Initiative focuses on practical deployment of AI for improving emergency coordination and intervention efficiency across the region.
ChatGPT adds interactive visual explanations for math and science, enabling students to explore formulas, variables, and concepts in real time.
OpenAI deploys a custom ChatGPT instance on GenAI.mil, a secure platform for U.S. defense teams. The service includes safety controls and compliance measures tailored to government requirements. No technical details or timeline provided.
TRUSTBANK and Recursive built Choice AI using OpenAI models to deliver personalized conversational recommendations for Furusato Nozei gifts. The AI agent streamlines regional Japanese donation discovery. Commercial deployment of conversational AI.
Podium built "Jerry," a GPT-5-powered AI agent deployed to 10,000+ SMBs for customer service automation. Claims 300% growth for users. Use case: Main Street businesses automating customer interactions via AI agents.
OpenAI launches its first certification courses and AI Foundations program to build practical AI skills. Training initiative aimed at improving career opportunities and preparing workers for the evolving job market.
OpenAI appoints Denise Dresser as Chief Revenue Officer overseeing global revenue strategy, enterprise expansion, and customer success to accelerate AI adoption across businesses.
OpenAI reports accelerating enterprise AI adoption in 2025 with deeper integration and measurable productivity gains across industries. No specific metrics disclosed in excerpt.
Mirakl integrates ChatGPT Enterprise and AI agents to speed up documentation, enhance customer support, and build agent-native commerce through Mirakl Nexus platform.
OpenAI partners with independent experts to evaluate frontier AI systems. Third-party testing strengthens safety, validates safeguards, and increases transparency in assessing model capabilities and risks.
OpenAI partners with Japan's Digital Agency to deploy generative AI in public services, strengthen international AI governance, and promote safe, trustworthy AI adoption globally.
OpenAI deploys an internal research assistant to analyze millions of support tickets, accelerate insight extraction, and scale investigation capacity across the organization.
OpenAI joins the EU Code of Practice for responsible AI. The company commits to collaborating with European governments on innovation, infrastructure, and economic growth.
OpenAI announces opening a Germany office to strengthen European presence and accelerate model adoption. The expansion reflects OpenAI's geographic growth strategy amid increasing competition from European and US players.
OpenAI revisits its findings on sycophancy—the tendency of models to tell users what they want to hear. The post details gaps in their initial study, methodological errors, and future changes to better measure this bias. No specific model names or metrics provided in the excerpt.
Google DeepMind expands access to Music AI Sandbox with new features for music professionals. The tool enables exploration of generative AI capabilities applied to music creation.
OpenAI submits recommendations to strengthen U.S. AI leadership based on its Economic Blueprint. Proposals address infrastructure, regulation, and public investment to maintain competitiveness against China and the EU.
Genmab deploys ChatGPT Enterprise to accelerate drug discovery and data analysis. The partnership leverages OpenAI's security and privacy commitments.