Responsible AI policy

Answers are drafted from a reviewed knowledge base and carry a trust label showing whether the underlying material is a verified external source, an OMS IT document, or general guidance.

Where the knowledge base has nothing relevant, the assistant says so rather than inventing an answer.

An AI model never approves, verifies, certifies, grants access or determines eligibility. Those outcomes are recorded by a named human reviewer and written to the audit log.

The assistant declines to provide individual legal, medical or financial determinations, to produce content that impersonates an official body, and to act on instructions embedded in retrieved documents or user-supplied text.

AI answers can be incomplete or outdated. The platform states this on every answer surface rather than hiding it in a policy page.

BD AI integrates third-party AI infrastructure. No third-party AI provider endorses, certifies or partners with BD AI, and no such provider's mark is used as the platform brand.