AI decisions and responsibility
TracePoint is decision-control infrastructure. The reference may capture proposed actions, purpose and context, authority, policy results, risk signals, approvals, dispositions, and audit evidence. It is not the underlying model, system of record, legal adviser, medical professional, or final business decision-maker.
AI outputs and risk signals may be incomplete or incorrect. Evaluation participants must establish appropriate testing, monitoring, validation, appeal, and human oversight for consequential uses.