AI news for builders and product teamsUpdated Nov 17, 2026, 13:01 UTC
Google Research
First-party releases and research from Google Research. Headlines and excerpts link to the original articles.
Latest stories
Newest first
Google Research says Antigravity's /teamwork multi-agent system, built on Gemini 4 models, has helped human experts resolve long-open problems in math, chemistry and systems software, and the tools are now available externally.

Google Research and Google DeepMind have launched Language Explorer, an interactive web platform built on the open-source LinguaMeta dataset that lets researchers, educators and language enthusiasts visualize linguistic diversity across thousands of languages. It is also installed as an interactive exhibit at Berlin's Museum of Communication.

Google Research published a Nature study describing AI-PPG-ACC-HTN, a smartwatch system that uses PPG and accelerometer data to flag persistent hypertension without cuff calibration. Validated against 24-hour ambulatory monitoring in 196 participants, it reached 65.8% sensitivity and 90.0% specificity for initial screening.

Google Research has proposed a "sleep framework" for continual learning in which LLMs alternate between active interaction and offline consolidation, expanding Mixture-of-Experts parameters to absorb new facts without catastrophic forgetting. It reports near-zero degradation in continual translation and near-perfect scaling to 10M tokens on BABILong.

A prospective multi-center study published in JAMA Network Open found that Google's AI model for gestational age assessment from "blind sweep" ultrasounds matched expert sonographers, with a mean absolute error of 4.18 days versus 4.48 days for standard biometry.

Google Research described three projects aimed at letting AI agents run real wet-lab experiments, including a digital-twin scaffolding for liquid-handling robots, closed-loop aptamer design, and de novo protein binders with single-digit nanomolar affinity for the cancer target FAP.

Google Research released a technical report on its Geospatial Reasoning agent, which links Gemini to Earth AI foundation models and a planning stage to run multi-step spatial analysis, and is expanding experimental access through Ask Google Earth and an Agent-to-Agent API.

At its Research@ event in Mountain View, Google Research announced ten updates spanning planetary intelligence, medical AI, scientific tooling and edge hardware. Highlights include expanded early access to Google Earth AI, public availability of the Deep Review document checker, and a next-gen low-power Edge AI chip from Synaptics.

Google Research has launched Deep Review, an agentic AI system for technical document verification, worldwide. The tool grew out of its Paper Assistant Tool, piloted at four computer science conferences where over 16,000 manuscripts received opt-in feedback.

A three-month randomized trial with 133 patent attorneys found AI drafting tools improved work quality for all lawyers, but only senior lawyers (7+ years) showed gains in unassisted judgment; juniors saw no average skill improvement and their scores split into more strong and more low results.

Google Research presents five partner-led case studies using its Earth AI Population Dynamics Foundation Model (PDFM) as a geospatial foundation model for public health. PDFM's location embeddings matched or improved conventional inputs across vaccination, cardiovascular disease, dengue, postpartum depression, and cholera tasks.

Google Research published a workshop report on agentic privacy and security, produced with more than 50 academic and industry participants at the Google Contextual Agent Privacy and Security Workshop held in late 2025 in New York City. The report outlines open research directions across system, model, user, and ecosystem levels.

Google Research announced a next-generation Federated Learning system that uses Trusted Execution Environments (TEEs) to provide externally verifiable privacy guarantees while moving computation to the server for faster training and better accuracy. Gboard has adopted the system for English and Japanese next-word prediction models.