AI Ethics Framework

Fairness, transparency, accountability, privacy, safety, and human agency. Every AI feature we ship is assessed against these six principles before release.

We identify, assess, and mitigate AI risks through a structured review aligned with the NIST AI RMF, covering intended use, misuse potential, and failure modes.

We test AI behavior for biased outcomes and correct issues we find. AI features must not be used to discriminate unlawfully, and we design defaults to avoid it.

People stay in control. Users can review, edit, and override AI output, pause agents, and set boundaries on what agents may do autonomously.

We tell users when they are interacting with AI, document what our AI features do, and give customers the tools to disclose AI use to their own customers.

AI features follow the same data rules as the rest of the platform: data minimization, per-tenant isolation, and no training on customer content. See the Privacy Policy and AI Data Training Policy.

We test agents against adversarial and edge-case inputs, apply content moderation to generated output, and work to reduce fabricated answers in high-stakes flows.

AI-driven actions are logged so behavior can be audited and explained after the fact. Responsibility for the platform's AI rests with us, not with the model providers we use.

We monitor AI behavior in production, act on user reports, and update this framework as the technology and the law evolve. Feedback: [email protected].