Back to Google (Gemini)All frontier AI labs
Ethics and safety policies
Google (Gemini): what they say, and what they leave out
Every claim on this page comes from documents Google (Gemini) published itself, linked below so you can check me. Checked September 19, 2026.
A policy is not an audit
Almost every document linked on this page was written by the company it governs, and graded by that same company. A long policy list does not make a lab more ethical than a short one. This page keeps three things apart: what the lab has committed to in writing, what its documents do not cover or where they have been credibly criticized, and — clearly labeled at the end — my own read.
Their stated position
Google publishes AI principles and a Frontier Safety Framework, and runs responsibility reviews inside its product organizations.
It reports on AI responsibility annually and ships model cards for the Gemini and Gemma families.
The documents themselves
Read these rather than this summary if a decision depends on it.
- Google AI Principles
Its stated principles for AI development and application.
- Frontier Safety Framework
Critical capability levels and the mitigations required at each.
- Responsible AI progress report
Its own annual account of governance, evaluations and safeguards.
- Prohibited use policy
Forbidden applications of its generative products.
What they have committed to
Stated in their own published policies, in writing.
- — Publishes model cards and technical reports for Gemini and open-weight Gemma releases.
- — Commits to capability evaluations before frontier deployments under the Frontier Safety Framework.
- — Publishes a recurring public report on responsibility work rather than a one-off statement.
What those documents don't cover
Plain absences, and criticism that has been documented publicly — not speculation.
- — In early 2025 Google removed its previous commitment not to build AI for weapons or surveillance from the principles page — a substantive reversal, not a rewording.
- — It dismissed members of its own ethics leadership in 2020-21 after internal disagreement over publication, which remains the most cited case of a lab suppressing its own researchers.
- — Model cards for Gemini releases have been criticized as thin and sometimes late relative to launch.
- — Training data for Gemini is undisclosed.
My read
This section is my opinion, clearly separated from everything above. Disagree with it freely.
Google has the most institutional machinery of any lab here and the clearest documented case of overruling it. Read the current principles page with the 2025 revision in mind.