How Mobility Data Can Give Language Models a Better Sense of Place
Google Research explores how movement patterns may help models understand locations beyond text-based descriptions.
AI & ML
Frontier models, research results, agents, and developer tooling.
OpenAI Bans Russian AI Influence Campaign Posing as Think Tank
OpenAI disrupted a covert Russian operation that used AI to promote a fake think tank and a sovereignty index.
Granite 4.2 LLMs: How IBM Built Its New Model Family
A Hugging Face technical post details the architecture and training choices behind IBM's Granite 4.2 large language models.
AI-coded open-source Adobe clones aim for full parity
A developer says AI-assisted reverse engineering produced seven open-source Adobe alternatives, though skeptics note the apps are far from complete.
OpenAI and Atlassian expand partnership to embed GPT-6 agents in Rovo
The expanded deal brings OpenAI frontier models into Atlassian's Rovo and developer workflows, grounding them in enterprise context.
Reka's Rho-1: A 19B Omni-Reasoning Model for Text, Video, and Robot Actions
Reka's Rho-1 is a single 19B model that handles text, images, video, and robot actions, with a distilled variant generating short clips in about a second.
Probing Text Embeddings: Interpreting Scikit-LLM Spaces with UMAP and SHAP
A practical guide to making LLM-based text embeddings more interpretable through probing classifiers, dimensionality reduction, and model explanations.
OpenAI backs Thai startups, widens Brazil footprint
OpenAI announces an eight-week accelerator for Thai health and education startups and a broader push into Brazil's developer and business ecosystem.
Open ASR Leaderboard Adds First Global South Language
The addition signals a move toward more inclusive speech recognition technology.
Gemini Omni 1.1 Flash puts developer control first
Google DeepMind's latest model update emphasizes giving developers more control over AI outputs.
Google's Earth AI automates global climate and weather models
A new framework from Google Research uses machine learning to automate and streamline the construction of planetary-scale Earth system models.
ChatGPT Ads Hits $1B Run Rate as OpenAI Expands Globally
OpenAI says its ChatGPT advertising business has reached a $1 billion annualized revenue run rate and is expanding internationally.
DeepMind Pilots Double-Blind Evaluations to Reduce AI Assessment Bias
Google DeepMind says it is running the world's first double-blind AI evaluations, an attempt to make model comparisons less biased and more rigorous.
OpenAI to end Cursor model supply after SpaceX acquisition
OpenAI says it is winding down its contract to supply models to Cursor because of the code tool's acquisition by SpaceX.
Critical-Thinking Training Boosts ChatGPT Student Outcomes
A randomized study of over 1,000 students links critical-thinking instruction to better answers and greater originality when using ChatGPT.
Google Adds AI Travel Booking and Fare Tracking to Search
Google's AI Mode in Search now supports hotel booking, airfare tracking, and rewards visibility, expanding its travel planning capabilities.
Gemini 3.5 Transcribe brings smarter speech-to-text
Google DeepMind's new Gemini 3.5 Transcribe model promises more intelligent speech-to-text transcription.
OpenAI Shares Findings from Hugging Face Incident and Security Overhaul
OpenAI's post-incident report details the breach's impact and outlines new safeguards for AI model security and alignment.