Transparency
Every AI interaction with a natural person discloses that it is AI, in the language and form required by the applicable jurisdiction. Disclosure is cryptographically attested per call and stored in an audit-ready evidence chain.
Responsible AI Commitment
Responsible AI at Audact.
We treat Responsible AI as operational practice, not as a principles document. Every commitment below corresponds to a control in the platform.
Human oversight
Audact deployments require a named human operator with the authority and tooling to intervene, pause, or terminate any conversation. Escalation paths to a human are configured per use case before go-live.
Robustness
Pre-dispatch compliance checks block calls that would breach disclosure or jurisdictional rules. Brand-governance overlays validate pricing, claims, and commercial commitments in real time. Incident detection runs continuously against the evidence chain.
Fairness
Audact does not develop user-facing dialogue policy; we govern it. Customers configuring deployments take FRIA-style impact assessments through the platform, with structured prompts for protected-characteristic risks under the AI Act and national equality law.
Aligned standards
Audact aligns with the EU AI Act, ISO/IEC 42001 AI Management System (NEN-track), NIST AI Risk Management Framework, and the OECD AI Principles. Where these standards conflict, EU AI Act provisions take precedence.