Google Research announced 10 research updates at its Research@ event in Mountain View on October 14, 2026, covering public health, scientific research, document review, edge hardware, business forecasting, medicine, disaster prediction, education and team prototyping. The company said the projects aim to put research capabilities in the hands of doctors, scientists, developers, public health officials and others.

Google Earth AI is expanding early access with three ways to use experimental capabilities: deeper geospatial analysis in Google Earth, risk assessments and scenario forecasts through a planetary prediction engine, and geospatial agent skills for custom workflows through an experimental agent-to-agent API. Google is also integrating Dataset Search, which queries more than 100 million public datasets, and TimesFM forecasting. The system draws on more than 70 geospatial tools, including planetary foundation models, and its results are validated by empirical benchmarks in an Earth AI technical report. Google.org is providing $10 million to crisis resilience and public health organizations, including $2 million to the Milken Institute for infrastructure work.

A multi-agent orchestrator helps scientists reproduce earlier methods and analyses, then apply them to new problems. In one example, researchers used a method for inverse design in fluid mechanics to develop urinary catheter designs that simulations found better at preventing bacterial infections. Google also said its Empirical Research Assistance tool generates and tests thousands of code variations in parallel, while Co-Scientist has helped Stanford researchers identify existing medicines for liver fibrosis. Deep Review, an agentic system for checking complex documents, has been used by Google researchers and external authors to review more than 16,000 research papers submitted to leading computer science conferences. Google is expanding public access so people can check documents including legal agreements and financial reports.

Google said Synaptics will reveal a co-designed, low-power Edge AI chip next week, using Google's open-source intellectual property. It is intended for smart glasses, watches, connected cameras, streamers, and industrial vision and voice systems. TimesFM and TabFM, models for time-series data and structured tables, make predictions without training and rank at the top of public benchmarks in their fields. They are available to Cloud BigQuery users, with integration into Gemini for business planned.

MedGemma was published in Nature Medicine last week and has been downloaded more than 10 million times by more than 65,000 innovators. Developers have adapted it for local health needs, including clinic triage and screening for eye disease and cervical cancer. Google is researching tool use and additional medical data types, including mammography, ultrasound and electrocardiograms. In a study published in The Lancet, 98 patients consulted AMIE before urgent care visits; clinicians said its summaries helped them prepare in 75% of cases. Google is also pursuing real-time video consultations and a nationwide randomized study with Included Health.

Google said its global hydrologic model provides river forecasts up to 7 days ahead for more than 2 billion people in over 150 countries. Groundsource used Gemini to turn flood news reports into a dataset of over 2.6 million flood events, helping train a model that predicts urban flash-flood likelihood up to 24 hours ahead; the dataset is publicly available. Google also described teacher-vetted learning interactives, with a sample library of over 30 AI-generated STEM simulations, and Play with Putty, a shared prototyping environment with real-time collaborative editing. Its waitlist is open on Google Labs.

Google Research announced 10 research updates at its Research@ event in Mountain View on October 14, 2026, covering public health, scientific research, document review, edge hardware, business forecasting, medicine, disaster prediction, education and team prototyping. The company said the projects aim to put research capabilities in the hands of doctors, scientists, developers, public health officials and others.

Google Earth AI is expanding early access with three ways to use experimental capabilities: deeper geospatial analysis in Google Earth, risk assessments and scenario forecasts through a planetary prediction engine, and geospatial agent skills for custom workflows through an experimental agent-to-agent API. Google is also integrating Dataset Search, which queries more than 100 million public datasets, and TimesFM forecasting. The system draws on more than 70 geospatial tools, including planetary foundation models, and its results are validated by empirical benchmarks in an Earth AI technical report. Google.org is providing $10 million to crisis resilience and public health organizations, including $2 million to the Milken Institute for infrastructure work.

A multi-agent orchestrator helps scientists reproduce earlier methods and analyses, then apply them to new problems. In one example, researchers used a method for inverse design in fluid mechanics to develop urinary catheter designs that simulations found better at preventing bacterial infections. Google also said its Empirical Research Assistance tool generates and tests thousands of code variations in parallel, while Co-Scientist has helped Stanford researchers identify existing medicines for liver fibrosis. Deep Review, an agentic system for checking complex documents, has been used by Google researchers and external authors to review more than 16,000 research papers submitted to leading computer science conferences. Google is expanding public access so people can check documents including legal agreements and financial reports.

Google said Synaptics will reveal a co-designed, low-power Edge AI chip next week, using Google's open-source intellectual property. It is intended for smart glasses, watches, connected cameras, streamers, and industrial vision and voice systems. TimesFM and TabFM, models for time-series data and structured tables, make predictions without training and rank at the top of public benchmarks in their fields. They are available to Cloud BigQuery users, with integration into Gemini for business planned.

MedGemma was published in Nature Medicine last week and has been downloaded more than 10 million times by more than 65,000 innovators. Developers have adapted it for local health needs, including clinic triage and screening for eye disease and cervical cancer. Google is researching tool use and additional medical data types, including mammography, ultrasound and electrocardiograms. In a study published in The Lancet, 98 patients consulted AMIE before urgent care visits; clinicians said its summaries helped them prepare in 75% of cases. Google is also pursuing real-time video consultations and a nationwide randomized study with Included Health.

Google said its global hydrologic model provides river forecasts up to 7 days ahead for more than 2 billion people in over 150 countries. Groundsource used Gemini to turn flood news reports into a dataset of over 2.6 million flood events, helping train a model that predicts urban flash-flood likelihood up to 24 hours ahead; the dataset is publicly available. Google also described teacher-vetted learning interactives, with a sample library of over 30 AI-generated STEM simulations, and Play with Putty, a shared prototyping environment with real-time collaborative editing. Its waitlist is open on Google Labs.