New: Evidence-linked scoring is here — every assessment cites the exact moment it's based on. See how →

Hireflo

Hireflo · Legal

Responsible AI Principles

Last updated August 24, 2026. Working draft for the product as shipped — have counsel review before relying on it in a customer contract.

  1. Evidence-based only. No credit without a cited transcript span.
  2. No emotion, face, or accent analysis. Content of answers, not how someone looks or sounds.
  3. Bias testing on purpose. We perturbation-test scoring and will publish third-party audits when we have them.
  4. Human oversight. AI recommends. People decide. Auto-reject is not offered.
  5. Candidate transparency. Invite pages say that AI is in the room, that the session is recorded, and how to access or delete data.
  6. Data minimization. We keep what the interview needs, not a shadow profile of the person.
  7. Override is first-class. A reviewer can change any criterion; the tape stays.