Responsible AI
AI that respects users
Our AI features are built with safeguards that prioritize accuracy, fairness, transparency, and user agency.
Principles
Commitments behind every AI feature
- Human-in-the-loop oversight for consequential decisions
- Clear disclosure when AI is in use
- Transparency about data used to power features
- Bias and fairness considerations during design
- Mechanisms for user feedback and correction
- Continuous evaluation of model behavior
Controls
User control and consent
- Opt-in pathways for sensitive AI capabilities
- Granular control over AI-generated suggestions
- Ability to review and edit AI outputs before use
- Clear pathways to disable AI features where applicable
- Documentation of feature limitations
- Respect for user-provided context boundaries
Safety
Abuse prevention and quality
- Guardrails against unsafe or harmful outputs
- Content filtering for sensitive categories
- Monitoring for misuse patterns
- Vendor diligence on AI providers
- Privacy-respecting prompts and logs
- Regular review of policies and prompts