Public information isn't just useful for AI training; it's important for a healthy society. Google's AI models are trained mainly on data from the open internet. This includes things like web pages, blog posts, and code — often with text, images, audio, and video. We don't access data from private databases or pages that require a login, unless the website owner allows it.

Generative AI and Privacy
A working paper of our insights on Generative AI and data privacy, and recommendations for applying long-standing protection principles.

Responsible Development And Use of Privacy Enhancing Technologies: How Governments Can Help
A working paper of our latest insights on Privacy Enhancing Technologies (PETs) and Google's approach to developing and using PETs in the products we build for our users and customers.

There's been much discussion about AI responsibility across the industry and civil society, but a widely accepted definition has remained elusive. AI responsibility has come to be associated with avoiding risks, but we actually see it as having two aspects: mitigating complexities and risks, as well as helping to improve people's lives and addressing social and scientific challenges.

To us, responsible deployment is not a single event, but an end-to-end AI Responsibility Lifecycle that integrates safety, security, and ethical considerations from the earliest research stages through to post-launch monitoring.

A shared agenda for responsible AI progress
Our core principles can help guide policy that promote progress while reducing risks of abuse.

Our 2026 Responsible AI Progress Report 
Sharing how we’re applying our AI Principles to the development of our products and research

End-to-end responsibility
Our AI Responsibility Lifecycle is a four-phase process that guides responsible AI development at Google.

Strengthening our Frontier Safety Framework
Our most comprehensive approach yet to identifying and mitigating severe risks from advanced AI models.

At Google, we operate as a good grid citizen, paying for 100% of the energy we use and the costs directly associated with our growth. Our approach to protect ratepayers also involves bringing new power to the grid, investing in innovative energy solutions, and creating local and long-term jobs in the communities we call home.

The Capacity Commitment Framework 
Our framework for helping to accelerate economic growth while protecting other ratepayers and ensuring we pay for the electricity to serve our operations.

Powering a New Era of American Innovation
15 policy opportunities to increase the capacity of the existing U.S. energy system.

Pennsylvania Energy & Innovation Summit
Our plans to invest in increasing America’s energy abundance and essential AI skills development in the Commonwealth and across the nation.

Artificial General Intelligence (AGI) is AI that’s at least as capable as humans at most cognitive tasks. Integrated with agentic capabilities, AGI could supercharge AI to understand, reason, plan, and execute actions autonomously. Such technological advancement will provide society with invaluable tools to address critical global challenges, including drug discovery, economic growth and climate change. To realize the benefits of this transformational technology, we are putting safety and responsibility at the heart of its development.

Levels of AGI framework
A perspective from Google DeepMind on classifying the capabilities of advanced AI systems.


An Approach to Technical AGI Safety and Security
Insights on how Google DeepMind is taking a responsible path to AGI.

Distributional AGI safety
Frontier AI research on building a safety net that grows and scales automatically with increasingly complex multi-agent systems.

AI agents represent an evolution in AI innovation. Agents help people and businesses get things done by taking action on their behalf and under their supervision. By working across apps and datasets that the user chooses to share with the agent, these tools enable people and businesses to spend more time on what matters most to them. Like Google's other AI tools, AI agents are built to be secure by design and with users in control. This ensures that the technology works for you.

Google's Approach for Secure AI Agents
As part of our ongoing efforts to define best practices for secure AI systems, we’re sharing our aspirational framework for secure AI agents.

To ensure Gemini Robotics benefits humanity, we’ve taken a comprehensive approach to safety, from practical safeguards to collaborations with experts, policymakers, and our Responsibility and Safety Council.

Responsibly advancing AI and robotics
Google DeepMind has developed broad and rigorous safety frameworks so Gemini-controlled robots can be used responsibly in real-life environments.