Our commitment to responsibility
We design for trust by ensuring AI capabilities are governed, deployed, and scaled in ways that are ethical, transparent, and aligned with human-centred values. To do this, we align our policies, technical safeguards, cross-functional oversight, and stakeholder engagement to four core principles. These four principles amplify the values and commitments in our Code of Conduct, Corporate Responsibility and Sustainability initiatives, and our contractual and regulatory obligations.
- Fairness and inclusiveness
We strive to create AI that promotes inclusiveness and equitable outcomes, avoids creating or reinforcing unfair bias, enables compliance, and allows managers to be present with their people.
- Transparency and interpretability
AI should support human judgment, not replace it. That’s why we disclose when customers are interacting with AI, what the AI is doing, and what its limits are.
- Privacy and security
We treat privacy and data stewardship as design requirements. Safeguards are applied to protect customer data and our models from unauthorized access, disclosure, and manipulation. Learn more about our privacy practices and the technical and organizational safeguards in place to ensure the security of our technology.
- Reliability and safety
We believe AI should be reliable, appropriate, and safe for its intended use. Our AI capabilities are designed to support consistent performance and reduce the risk of unintended or harmful outcomes.
Trust in practice
Responsible AI is a core commitment across all our teams, including legal, privacy, security, product, and engineering. Our shared values help us translate regulatory requirements and customer expectations into auditable controls and in-product experiences.
-
Governance by design
Our teams intentionally embed responsible AI into the product development and deployment lifecycle through governance checks, technical safeguards, cross-functional oversight, and ongoing monitoring.
-
Transparency by design
We explain when, how, and why AI is used and what safeguards exist. We provide customers with relevant information, including appropriate uses, limitations, evaluation results, and human-in-the-loop controls.
-
Reliability and safety
We continuously train, test, and monitor our models to reduce drift and improve trustworthiness over time.
-
Data use and minimization
We explain how data is used in connection with our services. We don’t use customer data or personal data for user profiling, advertising or similar commercial purposes, or to sell or share personal data. We may process personal data as needed to deliver, support, and improve those services. We may also create aggregated, non-personal data from de-identified or pseudonymized information, such as usage logs, to analyze performance, improve functionality, and develop new features. For more information see the enhanced data use disclosures in our DPA.
-
Performance monitoring
We test and monitor AI models for accuracy, resilience, and security. This involves collecting telemetry data, including usage statistics and limited user information such as mobile number, email address, and IP address, as needed to operate, secure, support, and improve these solutions.
-
Applying security processes
Layered security controls support our AI-powered solutions and help protect against unauthorized access, data exposure, and service disruption. We apply redaction, masking, and tokenization to user requests and model responses when applicable to reduce the exposure of sensitive information. We also protect data used for telemetry, monitoring, and performance analysis, and deploy data and models within isolated private cloud environments to further limit exposure.
Responsible AI in our products
Responsible AI is part of how we build products. We embed it into product development and deployment through governance, technical safeguards, cross-functional oversight, and ongoing monitoring. To learn more, read our Global Impact Report or contact us.