- 1.1 No contracts with partners accused of human rights violations, unless an independent third-party assessment, publicly disclosed, verifies the technology will not be used to facilitate such violations. This responsibility follows from the UN Guiding Principles on Business and Human Rights, which Alphabet has long committed to in its own statement on human rights. The United Nations, Amnesty International, and Alphabet shareholders have all identified contracts upon which the UNGPs should compel Alphabet to act.
1.1 No contracts with partners accused of human rights violations, unless an independent third-party assessment, publicly disclosed, verifies the technology will not be used to facilitate such violations.
This responsibility follows from the UN Guiding Principles on Business and Human Rights, which Alphabet has long committed to in its own statement on human rights. The United Nations, Amnesty International, and Alphabet shareholders have all identified contracts upon which the UNGPs should compel Alphabet to act.
- 1.2 Restore red lines on harmful applications (surveillance, weapons). We will not pursue weapons, technologies or contracts whose principal purpose, implementation or impact causes harm or directly facilitates injury to people, or gather or use information for surveillance violating internationally accepted norms.
1.2 Restore red lines on harmful applications (surveillance, weapons).
We will not pursue weapons, technologies or contracts whose principal purpose, implementation or impact causes harm or directly facilitates injury to people, or gather or use information for surveillance violating internationally accepted norms.
- 1.3 AI principles in contract terms. AI principles must be a non-negotiable part of any contract and apply equally whether we build the whole system or just supply an AI component in someone else's pipeline.
- 1.4 AI principles must be durable, with a transparent, participatory process to change them.
- 1.5 AI principles must be enforced and monitored, with independent oversight. AI principles must be written in a way that makes them clear, specific, understandable, and enforceable. Google should then apply robust monitoring mechanisms that can detect almost any usage of our systems that violates the principles, and monitoring capabilities should be preserved in the deployment arrangements. Independent oversight should complement internal monitoring and enforcement, so that a party other than Google can verify principles are upheld in practice. If any violation is found, Google should immediately apply effective enforcement measures that cause this misuse to cease.
1.5 AI principles must be enforced and monitored, with independent oversight.
AI principles must be written in a way that makes them clear, specific, understandable, and enforceable. Google should then apply robust monitoring mechanisms that can detect almost any usage of our systems that violates the principles, and monitoring capabilities should be preserved in the deployment arrangements.
Independent oversight should complement internal monitoring and enforcement, so that a party other than Google can verify principles are upheld in practice. If any violation is found, Google should immediately apply effective enforcement measures that cause this misuse to cease.
- 1.6 Google must create accountability for violations, with real consequences. Within the company, Google must lay out meaningful, proportionate consequences for workers who take actions that violate our AI principles. These consequences must apply to any Googler at any level, without exceptions.
1.6 Google must create accountability for violations, with real consequences.
Within the company, Google must lay out meaningful, proportionate consequences for workers who take actions that violate our AI principles. These consequences must apply to any Googler at any level, without exceptions.